1. Data Cleaning

Importing, Analyzing the Data and set visualization styles

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
sns.set(style="whitegrid")

Analyzing the data

Load the Bumble Dataset

bumble_path = r"C:\Users\HP\OneDrive\Documents\NextLeap Projects\Python\bumble.csv"
df = pd.read_csv(bumble_path)

Display dataset structure

df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 59946 entries, 0 to 59945
Data columns (total 17 columns):
 #   Column       Non-Null Count  Dtype  
---  ------       --------------  -----  
 0   age          59946 non-null  int64  
 1   status       59946 non-null  object 
 2   gender       59946 non-null  object 
 3   body_type    54650 non-null  object 
 4   diet         35551 non-null  object 
 5   drinks       56961 non-null  object 
 6   education    53318 non-null  object 
 7   ethnicity    54266 non-null  object 
 8   height       59943 non-null  float64
 9   income       59946 non-null  int64  
 10  job          51748 non-null  object 
 11  last_online  59946 non-null  object 
 12  location     59946 non-null  object 
 13  pets         40025 non-null  object 
 14  religion     39720 non-null  object 
 15  sign         48890 non-null  object 
 16  speaks       59896 non-null  object 
dtypes: float64(1), int64(2), object(14)
memory usage: 7.8+ MB

Display first five rows

df.head()

	age	status	gender	body_type	diet	drinks	education	ethnicity	height	income	job	last_online	location	pets	religion	sign	speaks
0	22	single	m	a little extra	strictly anything	socially	working on college/university	asian, white	75.0	-1	transportation	2012-06-28-20-30	south san francisco, california	likes dogs and likes cats	agnosticism and very serious about it	gemini	english
1	35	single	m	average	mostly other	often	working on space camp	white	70.0	80000	hospitality / travel	2012-06-29-21-41	oakland, california	likes dogs and likes cats	agnosticism but not too serious about it	cancer	english (fluently), spanish (poorly), french (...
2	38	available	m	thin	anything	socially	graduated from masters program	NaN	68.0	-1	NaN	2012-06-27-09-10	san francisco, california	has cats	NaN	pisces but it doesn’t matter	english, french, c++
3	23	single	m	thin	vegetarian	socially	working on college/university	white	71.0	20000	student	2012-06-28-14-22	berkeley, california	likes cats	NaN	pisces	english, german (poorly)
4	29	single	m	athletic	NaN	socially	graduated from college/university	asian, black, other	66.0	-1	artistic / musical / writer	2012-06-27-21-26	san francisco, california	likes dogs and likes cats	NaN	aquarius	english

Count of missing values per column

import pandas as pd  
bumble_path = r"C:\Users\HP\OneDrive\Documents\NextLeap Projects\Python\bumble.csv" 
df = pd.read_csv(bumble_path)
df.head()
missing_values = df.isnull().sum()
print("Missing Values in Each Column:")
print(missing_values)

Missing Values in Each Column:
age                0
status             0
gender             0
body_type       5296
diet           24395
drinks          2985
education       6628
ethnicity       5680
height             3
income             0
job             8198
last_online        0
location           0
pets           19921
religion       20226
sign           11056
speaks            50
dtype: int64

Percentage of Missing Values for All Columns

missing_values = df.isnull().sum()
missing_percentage = (df.isnull().sum() / len(df)) * 100
missing_data = pd.DataFrame({'Total Missing Values': missing_values, 'Percentage Missing': missing_percentage})
missing_data = missing_data.sort_values(by='Percentage Missing', ascending=False)
print("Missing Values Summary:")
print(missing_data)

Missing Values Summary:
             Total Missing Values  Percentage Missing
diet                        24395           40.694959
religion                    20226           33.740366
pets                        19921           33.231575
sign                        11056           18.443266
job                          8198           13.675641
education                    6628           11.056618
ethnicity                    5680            9.475194
body_type                    5296            8.834618
drinks                       2985            4.979482
speaks                         50            0.083408
height                          3            0.005005
last_online                     0            0.000000
location                        0            0.000000
income                          0            0.000000
status                          0            0.000000
gender                          0            0.000000
age                             0            0.000000

Findings: No columns should be dropped since the missing percentage is less than 50

df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 59946 entries, 0 to 59945
Data columns (total 17 columns):
 #   Column       Non-Null Count  Dtype  
---  ------       --------------  -----  
 0   age          59946 non-null  int64  
 1   status       59946 non-null  object 
 2   gender       59946 non-null  object 
 3   body_type    54650 non-null  object 
 4   diet         35551 non-null  object 
 5   drinks       56961 non-null  object 
 6   education    53318 non-null  object 
 7   ethnicity    54266 non-null  object 
 8   height       59943 non-null  float64
 9   income       59946 non-null  int64  
 10  job          51748 non-null  object 
 11  last_online  59946 non-null  object 
 12  location     59946 non-null  object 
 13  pets         40025 non-null  object 
 14  religion     39720 non-null  object 
 15  sign         48890 non-null  object 
 16  speaks       59896 non-null  object 
dtypes: float64(1), int64(2), object(14)
memory usage: 7.8+ MB

Handling missing numerical data (e.g., height, income) should be handled by imputing the median value of height and income for the corresponding category, such as gender, age group, or location.

# Impute 'height' based on gender
calculate_median_height = df.groupby(["gender"])["height"].transform("median")
df["height"] = df["height"].fillna(calculate_median_height)

# Impute 'income' based on gender and age group
calculate_median_income = df.groupby(["gender", "age"])["income"].transform("median")
df["income"] = df["income"].fillna(calculate_median_income)

# Print the calculated median values for verification
print("Calculated Median Heights:\n", calculate_median_height)
print("\nCalculated Median Income:\n", calculate_median_income)

Calculated Median Heights:
 0        70.0
1        70.0
2        70.0
3        70.0
4        70.0
         ... 
59941    65.0
59942    70.0
59943    70.0
59944    70.0
59945    70.0
Name: height, Length: 59946, dtype: float64

Calculated Median Income:
 0       -1.0
1       -1.0
2       -1.0
3       -1.0
4       -1.0
        ... 
59941   -1.0
59942   -1.0
59943   -1.0
59944   -1.0
59945   -1.0
Name: income, Length: 59946, dtype: float64

Explanation :

groupby(["gender"])["height"].transform("median")

• Groups the dataset by gender.
• Calculates the median height for each gender.
• Uses .transform("median") to return a column of median values matching the original dataset's shape.
• fillna(calculate_median_height)
• Fills missing height values using the previously calculated median height for each gender.

groupby(["gender", "age"])["income"].transform("median")

• Groups by both gender and age for more relevant median values.
• Ensures income values are filled in a contextually meaningful way (e.g., young men and older women may have different median incomes).
• Print statements for verification
• Displays the calculated median values to ensure accuracy before applying them.

Reason for doing:

• Prevents bias → Instead of filling with a global median, it fills based on meaningful categories.
• Preserves trends → Ensures the dataset remains representative of real-world distributions.
• Handles missing values efficiently → groupby().transform("median") applies the transformation without altering dataset structure.

To check are there any inconsistencies in the data types across columns (e.g., numerical data stored as strings)?

df.dtypes

age              int64
status          object
gender          object
body_type       object
diet            object
drinks          object
education       object
ethnicity       object
height         float64
income           int64
job             object
last_online     object
location        object
pets            object
religion        object
sign            object
speaks          object
dtype: object

• Used dtypes to check data types of each column.
• No inconsistencies in numerical columns, but last_online is stored as object and needs conversion to datetime.

To find which columns require conversion to numerical data types for proper analysis (e.g., income)?

• income and height are already stored as numerical (int64, float64 respectively).
• No further conversions are needed for numerical data types.

Does the last_online column need to be converted into a datetime format? What additional insights can be gained by analyzing this as a date field?

• Convert 'last_online' column to datetime format
• Display the updated column

df["last_online"] = pd.to_datetime(df["last_online"], format="%Y-%m-%d-%H-%M", errors="coerce")
df["last_online"]

0       2012-06-28 20:30:00
1       2012-06-29 21:41:00
2       2012-06-27 09:10:00
3       2012-06-28 14:22:00
4       2012-06-27 21:26:00
                ...        
59941   2012-06-12 21:47:00
59942   2012-06-29 11:01:00
59943   2012-06-27 23:37:00
59944   2012-06-23 13:01:00
59945   2012-06-29 00:42:00
Name: last_online, Length: 59946, dtype: datetime64[ns]

Findings:

• The last_online column was stored as an object type.
• We converted it into datetime64[ns] format using to_datetime().
• This allows for time-based analysis, such as determining inactive users or tracking user engagement trends.

Outliers

• 1. Are there any apparent outliers in numerical columns such as age, height, or income? What are the ranges of values in these columns?**

Display summary statistics to detect outliers

df.describe()

	age	height	income	last_online
count	59946.000000	59946.000000	59946.000000	59946
mean	32.340290	68.295282	20033.222534	2012-05-22 06:43:35.300770560
min	18.000000	1.000000	-1.000000	2011-06-27 01:52:00
25%	26.000000	66.000000	-1.000000	2012-05-29 20:37:15
50%	30.000000	68.000000	-1.000000	2012-06-27 14:30:00
75%	37.000000	71.000000	-1.000000	2012-06-30 01:09:00
max	110.000000	95.000000	1000000.000000	2012-07-01 08:57:00
std	9.452779	3.994738	97346.192104	NaN

Findings:

• This summary gives us key statistics like min, max, mean, and quartiles to detect extreme values.

Identify the range of numerical columns

age_range = (df['age'].min(), df['age'].max())
height_range = (df['height'].min(), df['height'].max())
income_range = (df['income'].min(), df['income'].max())

print(f"The age range is: {age_range}")
print(f"The height range is: {height_range}")
print(f"The income range is: {income_range}")

The age range is: (18, 110)
The height range is: (1.0, 95.0)
The income range is: (-1, 1000000)

Findings:

• The age range is (18, 110) → 110 is likely an outlier.
• The height range is (1.0, 95.0) → 1 inch is clearly incorrect.
• The income range is (-1, 1,000,000) → Negative values and extreme values exist.

Plotting Outliers Using Boxplots for all numerical columns

import matplotlib.pyplot as plt
import seaborn as sns

def plot_outliers(data, column):
    plt.figure(figsize=(8,5))
    sns.boxplot(y=data[column], palette='coolwarm')
    plt.title(f'Boxplot of {column}')
    plt.show()

plot_outliers(df, 'age')
plot_outliers(df, 'height')
plot_outliers(df, 'income')

C:\Program Files\KMSpico\temp\ipykernel_10508\4067286754.py:6: FutureWarning: 

Passing `palette` without assigning `hue` is deprecated and will be removed in v0.14.0. Assign the `x` variable to `hue` and set `legend=False` for the same effect.

  sns.boxplot(y=data[column], palette='coolwarm')

C:\Program Files\KMSpico\temp\ipykernel_10508\4067286754.py:6: FutureWarning: 

Passing `palette` without assigning `hue` is deprecated and will be removed in v0.14.0. Assign the `x` variable to `hue` and set `legend=False` for the same effect.

  sns.boxplot(y=data[column], palette='coolwarm')

C:\Program Files\KMSpico\temp\ipykernel_10508\4067286754.py:6: FutureWarning: 

Passing `palette` without assigning `hue` is deprecated and will be removed in v0.14.0. Assign the `x` variable to `hue` and set `legend=False` for the same effect.

  sns.boxplot(y=data[column], palette='coolwarm')

Findings:

• Boxplots visually confirm extreme outliers.
• The income column contains negative values, which need to be handled.

Any -1 values in numerical columns like income should be replaced with 0, as they may represent missing or invalid data.

• Check how many -1 values exist in income
• Replace -1 values with 0
• Verify replacement

(df['income'] == -1).value_counts() 
df.loc[df['income'] == -1, 'income'] = 0 
(df['income'] == -1).value_counts()

income
False    59946
Name: count, dtype: int64

Findings:

• Negative income values were replaced with 0 to reflect missing/unreported income.

Treating Outliers for Each Column Using the Middle 80% Approach

-- Handling Income Outliers

lower_bound_income = df['income'].quantile(0.10)
upper_bound_income = df['income'].quantile(0.90)

lower_outliers = df.loc[df['income'] < lower_bound_income, 'income']
upper_outliers = df.loc[df['income'] > upper_bound_income, 'income']

middle_80_income = df.loc[(df['income'] >= lower_bound_income) & (df['income'] <= upper_bound_income), 'income']

print(f"Lower bound income: {lower_bound_income}, Upper bound income: {upper_bound_income}")
print(f"Lower bound outliers range: {lower_outliers.min()} - {lower_outliers.max()}")
print(f"Upper bound outliers range: {upper_outliers.min()} - {upper_outliers.max()}")
print(f"Middle 80% mean: {middle_80_income.mean()}, Median: {middle_80_income.median()}")

Lower bound income: 0.0, Upper bound income: 50000.0
Lower bound outliers range: nan - nan
Upper bound outliers range: 60000 - 1000000
Middle 80% mean: 3297.01223769799, Median: 0.0

Findings:

• Outliers exist, but income outliers are expected for high earners.
• Conclusion: Outliers are kept as they reflect real-world variance.

-- Handling Age Outliers

lower_bound_age = df['age'].quantile(0.10)
upper_bound_age = df['age'].quantile(0.90)

lower_outliers = df.loc[df['age'] < lower_bound_age, 'age']
upper_outliers = df.loc[df['age'] > upper_bound_age, 'age']

middle_80_age = df.loc[(df['age'] >= lower_bound_age) & (df['age'] <= upper_bound_age), 'age']

print(f"Lower bound outliers range: {lower_outliers.min()} - {lower_outliers.max()}")
print(f"Upper bound outliers range: {upper_outliers.min()} - {upper_outliers.max()}")
print(f"Lower bound age: {lower_bound_age}, Upper bound age: {upper_bound_age}")

Lower bound outliers range: 18 - 22
Upper bound outliers range: 47 - 110
Lower bound age: 23.0, Upper bound age: 46.0

Findings:

• Users with age > 100 are likely errors.
• Conclusion: Possible capping at realistic values (e.g., 18-80 years).

-- Handling Height Outliers

lower_bound_height = df['height'].quantile(0.10)
upper_bound_height = df['height'].quantile(0.90)

lower_outliers = df.loc[df['height'] < lower_bound_height, 'height']
upper_outliers = df.loc[df['height'] > upper_bound_height, 'height']

print(f"Height lower bound: {lower_bound_height}, upper bound: {upper_bound_height}")
print(f"Lower outliers range: {lower_outliers.min()} - {lower_outliers.max()}")
print(f"Upper outliers range: {upper_outliers.min()} - {upper_outliers.max()}")

# Capping height values within a reasonable range
lower_bound = min(lower_bound_height, 48)  # Ensure no extreme low values
upper_bound = max(upper_bound_height, 90)  # Ensure no extreme tall values

df['height'] = df['height'].apply(lambda x: lower_bound if x < lower_bound else (upper_bound if x > upper_bound else x))

Height lower bound: 63.0, upper bound: 73.0
Lower outliers range: 1.0 - 62.0
Upper outliers range: 74.0 - 95.0

Findings:

• Heights below 48 inches or above 90 inches are unrealistic and need capping.
• Applied transformations to fix extreme outliers while preserving realistic data.

Conclusion

• Income outliers are kept, as high-income variations are expected.
• Age outliers above 100 are likely incorrect and require further review.
• Height outliers are capped within realistic human height ranges.

Missing Data Visualization

Create a heatmap to visualize missing values across the dataset. Which columns show consistent missing data patterns?

import matplotlib.pyplot as plt
import seaborn as sns
plt.figure(figsize=(12, 8))
sns.heatmap(df.isnull(), cbar=False, cmap="coolwarm")
plt.title('Heatmap of Missing Data')
plt.xlabel('Columns', fontsize=14)
plt.ylabel('Rows', fontsize=14)
plt.show()

Findings:

• The heatmap helps us identify patterns of missing data.
• Columns diet, religion, and pets show consistent missing data patterns.

Mode Imputation for Categorical Columns with Less Than 20% Missing Values

• Apply mode imputation to categorical columns with low missing values

for col in ['sign', 'job', 'education', 'ethnicity', 'body_type', 'drinks']:
    df[col] = df[col].fillna(df[col].mode()[0])

Findings:

• Mode imputation was applied to categorical columns with less than 20% missing values.
• This prevents unnecessary data loss while ensuring data integrity.

2. Data Processing

Binning and Grouping

Bin the age column into categories such as "18-25", "26-35", "36-45", and "46+" to create a new column, age_group.

• Define age bins and labels
• Apply binning to create a new age group column
• Display the first few rows to verify the new column

age_bins = [18, 25, 35, 45, float('inf')]
age_labels = ["18-25", "26-35", "36-45", "46+"]

df["age_group"] = pd.cut(df["age"], bins=age_bins, labels=age_labels)

df[["age", "age_group"]].head(10)

	age	age_group
0	22	18-25
1	35	26-35
2	38	36-45
3	23	18-25
4	29	26-35
5	29	26-35
6	32	26-35
7	31	26-35
8	24	18-25
9	37	36-45

Findings:

• Age is now categorized into meaningful groups for better analysis.
• Next step: Analyze the distribution of users across these age groups.

Count the number of users in each age group

age_distribution = df["age_group"].value_counts()
print(age_distribution)

age_group
26-35    28621
18-25    14145
36-45    10803
46+       6068
Name: count, dtype: int64

Findings:

• Majority of Bumble users fall within the 26-35 and 18-25 age groups.
• This suggests that the platform is primarily used by younger individuals.

Group income into categories like "Low Income," "Medium Income," and "High Income" based on meaningful thresholds (e.g., quartiles).

• Calculate income quartiles
• Assign income categories
• Display income category distribution

Q1 = df["income"].quantile(0.25)
Q2 = df["income"].quantile(0.50)
Q3 = df["income"].quantile(0.90)

df["income_category"] = np.where(df["income"] <= Q1, "Low Income",
                          np.where(df["income"] <= Q2, "Medium Income", "High Income"))

income_distribution = df["income_category"].value_counts()
print(income_distribution)

income_category
Low Income     48442
High Income    11504
Name: count, dtype: int64

Findings:

• Most users fall into the Low Income category, followed by High Income.
• This may indicate that many users report lower incomes or do not disclose higher earnings.

Derived Features

Create a new feature, profile_completeness, by calculating the percentage of non-missing values for each user profile.

• Calculate profile completeness percentage
• Display the first few rows to verify the new column

df["profile_completeness"] = df.notnull().sum(axis=1) / df.shape[1] * 100
print(df[["profile_completeness"]].head())

   profile_completeness
0            100.000000
1            100.000000
2             94.736842
3             94.736842
4             89.473684

Findings:

• Users with fewer missing values have a higher profile_completeness score.

Analyze profile completeness by relationship status

print(df.groupby("status")["profile_completeness"].mean())

status
available         94.973896
married           94.685908
seeing someone    94.777642
single            94.260014
unknown           91.578947
Name: profile_completeness, dtype: float64

Findings:

• Users who are single or seeing someone tend to have more complete profiles.
• The unknown category has the lowest completeness, indicating less engagement.
• Next step: Investigate whether profile completeness correlates with user activity.

Unit Conversion

Convert the height column from inches to centimeters using the conversion factor (1 inch = 2.54 cm).

• Convert height to centimeters
• Display the first few rows to verify conversion

df['height_cm'] = df['height'] * 2.54
print(df[["height", "height_cm"]].head())

   height  height_cm
0    75.0     190.50
1    70.0     177.80
2    68.0     172.72
3    71.0     180.34
4    66.0     167.64

Findings:

• Height is now in centimeters, making it easier to interpret for international users.
• Ensures standardization across different measurement systems.

3. Data Analysis

Demographic Analysis

What is the gender distribution across the platform? Are there any significant imbalances?

• Count the number of users by gender

gender_distribution = df['gender'].value_counts()
print("Gender distribution:")
print(gender_distribution)

Gender distribution:
gender
m    35829
f    24117
Name: count, dtype: int64

Findings:

• The platform has a higher number of male users compared to female users.
• This indicates a moderate gender imbalance, which may impact user experience and engagement.

What are the proportions of users in different status categories?

• Count the number of users by relationship status

status_distribution = df['status'].value_counts(normalize=True) * 100
print("Status distribution (Percentage):")
print(status_distribution)

Status distribution (Percentage):
status
single            92.911954
seeing someone     3.443099
available          3.111133
married            0.517132
unknown            0.016682
Name: proportion, dtype: float64

Findings:

• The majority of users (92.91%) are single, suggesting that Bumble mainly attracts single individuals looking for connections.
• Users identifying as seeing someone or available form a much smaller proportion.
• Marketing strategies should focus on singles seeking relationships.

How does status vary by gender?

• Analyze relationship status by gender

status_gender_distribution = df.groupby(['gender', 'status']).size() / len(df) * 100
print("Status distribution by gender (Percentage):")
print(status_gender_distribution)

Status distribution by gender (Percentage):
gender  status        
f       available          1.094318
        married            0.225203
        seeing someone     1.673173
        single            37.231842
        unknown            0.006673
m       available          2.016815
        married            0.291929
        seeing someone     1.769926
        single            55.680112
        unknown            0.010009
dtype: float64

Findings:

• 93% of men and 92% of women identify as single, reinforcing that the platform primarily serves singles.
• Gender-wise, there is no significant variation in relationship status distribution.

Correlation Analysis

What are the correlations between numerical columns (age, income, height, profile completeness)?

• Calculate correlation matrix

correlation_matrix = df[['age', 'income', 'height', 'profile_completeness']].corr()
print("Correlation Matrix:")
print(correlation_matrix)

Correlation Matrix:
                           age    income    height  profile_completeness
age                   1.000000 -0.001004 -0.022958              0.038296
income               -0.001004  1.000000  0.067056              0.060893
height               -0.022958  0.067056  1.000000             -0.011650
profile_completeness  0.038296  0.060893 -0.011650              1.000000

Findings:

• There are no strong correlations among age, income, height, and profile completeness.
• This suggests that these variables are largely independent.

How does age correlate with income?

• Compute correlation between age and income

age_income_corr = df['age'].corr(df['income'])
print(f"Correlation between age and income: {age_income_corr}")

Correlation between age and income: -0.0010038681910054018

Findings:

• Age and income have a very weak negative correlation (-0.001).
• Older users do not necessarily report higher incomes.

Diet and Lifestyle Analysis

How do dietary preferences distribute across the platform?

• Calculate diet distribution

diet_distribution = df['diet'].value_counts(normalize=True) * 100
print("Dietary Preference Distribution:")
print(diet_distribution)

Dietary Preference Distribution:
diet
mostly anything        46.651290
anything               17.391916
strictly anything      14.382155
mostly vegetarian       9.687491
mostly other            2.832550
strictly vegetarian     2.461253
vegetarian              1.876178
strictly other          1.271413
mostly vegan            0.950747
other                   0.931057
strictly vegan          0.641332
vegan                   0.382549
mostly kosher           0.241906
mostly halal            0.135017
strictly halal          0.050631
strictly kosher         0.050631
halal                   0.030941
kosher                  0.030941
Name: proportion, dtype: float64

Findings:

• 46% of users follow an 'anything' diet, while very few are vegetarian or vegan.
• Users with stricter diets (strictly vegan, kosher, halal) represent a small minority.

How do drinking habits vary across different diet categories?

• Group drinking habits by diet

drinking_by_diet = df.groupby('diet')['drinks'].value_counts(normalize=True) * 100
print("Drinking habits by diet:")
print(drinking_by_diet)

Drinking habits by diet:
diet        drinks     
anything    socially       76.419214
            often           9.283519
            rarely          8.248423
            not at all      4.771147
            very often      0.938056
                             ...    
vegetarian  rarely         10.044978
            often           8.845577
            not at all      5.547226
            very often      1.049475
            desperately     0.899550
Name: proportion, Length: 103, dtype: float64

Findings:

• Users with strict dietary preferences (e.g., strictly vegan) are less likely to drink alcohol.
• Social drinking is most common across all diet types.

Geographical Insights

Extract city and state information from location

• Split location column into city and state
• Display the first few rows

Steps:

• Ensure all locations have a comma before splitting
• Fill missing state values with "Unknown"
• Display the first few rows to verify

df[['city', 'state']] = df['location'].str.split(', ', expand=True, n=1)
df['state'] = df['state'].fillna('Unknown')
print(df[['location', 'city', 'state']].head(10))

                          location                 city       state
0  south san francisco, california  south san francisco  california
1              oakland, california              oakland  california
2        san francisco, california        san francisco  california
3             berkeley, california             berkeley  california
4        san francisco, california        san francisco  california
5        san francisco, california        san francisco  california
6        san francisco, california        san francisco  california
7        san francisco, california        san francisco  california
8    belvedere tiburon, california    belvedere tiburon  california
9            san mateo, california            san mateo  california

df[df['state'].isnull()]['location'].unique()

array([], dtype=object)

What are the top 5 cities and states with the highest number of users?

• Get top 5 cities and states

top_cities = df['city'].value_counts().head(5)
top_states = df['state'].value_counts().head(5)
print("Top 5 cities:")
print(top_cities)
print("Top 5 states:")
print(top_states)

Top 5 cities:
city
san francisco    31064
oakland           7214
berkeley          4212
san mateo         1331
palo alto         1064
Name: count, dtype: int64
Top 5 states:
state
california       59855
new york            17
illinois             8
massachusetts        5
texas                4
Name: count, dtype: int64

Findings:

• Most users are concentrated in California, New York, and Illinois.
• San Francisco is the most represented city.

Height Analysis

What is the average height of users across different gender categories?

• Calculate average height by gender

height_by_gender = df.groupby('gender')['height_cm'].mean()
print("Average height by gender:")
print(height_by_gender)

Average height by gender:
gender
f    165.378255
m    178.940090
Name: height_cm, dtype: float64

Findings:

• Men are taller (70.4 cm avg) than women (65.1 cm avg), which aligns with expected biological differences.

How does height vary by age group?

• Compute average height for each age group

height_by_age_group = df.groupby('age_group')['height_cm'].mean()
print("Average height by age group:")
print(height_by_age_group)

Average height by age group:
age_group
18-25    173.309345
26-35    173.763476
36-45    173.571604
46+      172.571401
Name: height_cm, dtype: float64

C:\Program Files\KMSpico\temp\ipykernel_10508\4240767852.py:1: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behavior or observed=True to adopt the future default and silence this warning.
  height_by_age_group = df.groupby('age_group')['height_cm'].mean()

Findings:

• Minimal height differences across age groups.
• Older users do not show a significant difference in height compared to younger users.

Income Analysis

What is the distribution of income across the platform (excluding 0 values)?

• Exclude zero-income values

income_distribution = df[df['income'] > 0]['income'].value_counts().sort_values(ascending=False)
print("Income distribution:")
print(income_distribution)

Income distribution:
income
20000      2952
100000     1621
80000      1111
30000      1048
40000      1005
50000       975
60000       736
70000       707
150000      631
1000000     521
250000      149
500000       48
Name: count, dtype: int64

Findings:

• The most common reported income is 20000 dollars
• Some users report very high incomes (more than 100,000 dollars), but they are a minority

How does income vary by age group and gender?

• Calculate average income by age group and gender

income_by_age_group = df[df['income'] > 0].groupby('age_group')['income'].mean()
income_by_gender = df[df['income'] > 0].groupby('gender')['income'].mean()
print("Income by age group:")
print(income_by_age_group)
print("Income by gender:")
print(income_by_gender)

Income by age group:
age_group
18-25    101760.808926
26-35    108869.783617
36-45    105937.645416
46+       91844.426624
Name: income, dtype: float64
Income by gender:
gender
f     86633.472535
m    110984.388035
Name: income, dtype: float64

C:\Program Files\KMSpico\temp\ipykernel_10508\1608165714.py:1: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behavior or observed=True to adopt the future default and silence this warning.
  income_by_age_group = df[df['income'] > 0].groupby('age_group')['income'].mean()

Findings:

• Income does not increase significantly with age.
• Men report slightly higher incomes than women.

4. Data Visualization

Age Distribution

• Plot a histogram of age with a vertical line indicating the mean age.
• Set plot style
• Create a histogram of age
• Calculate and plot mean age
• Labels and title

import matplotlib.pyplot as plt
import seaborn as sns

sns.set_style("whitegrid")

title_font = {'fontsize': 14, 'fontweight': 'bold'}
label_font = {'fontsize': 12}

plt.figure(figsize=(10, 6))
sns.histplot(df['age'], bins=20, kde=True, color='#3498db', edgecolor='black')

mean_age = df['age'].mean()
plt.axvline(mean_age, color='red', linestyle='--', label=f'Mean Age: {mean_age:.2f}')

plt.xlabel('Age', **label_font)
plt.ylabel('Frequency', **label_font)
plt.title('Distribution of Age with Mean Age Indicated', **title_font)
plt.legend()
plt.show()

Findings:

• Most users fall in the 20-30 age range.
• The distribution suggests that Bumble caters to a younger audience.
• Next Step: Compare age distribution across gender.

How does the age distribution differ by gender?

• Create a histogram of age distribution by gender

plt.figure(figsize=(10, 6))
ax = sns.histplot(df, x='age', hue='gender', palette='viridis', kde=True, bins=20, hue_order=df['gender'].unique())
if ax.get_legend() is not None:
    ax.legend(title="Gender", loc="upper right")

plt.xlabel('Age', fontsize=12)
plt.ylabel('Frequency', fontsize=12)
plt.title('Age Distribution by Gender', fontsize=14)
plt.show()

C:\Program Files\KMSpico\temp\ipykernel_10508\2309023062.py:4: UserWarning: No artists with labels found to put in legend.  Note that artists whose label start with an underscore are ignored when legend() is called with no argument.
  ax.legend(title="Gender", loc="upper right")

Findings:

• Both male and female users have a similar age distribution.
• There is a slightly higher number of male users in older age groups.

Income and Age

Scatterplot of income vs. age with trend line

• Scatterplot with regression trendline

plt.figure(figsize=(10, 6))
sns.regplot(data=df[df['income'] > 0], x='age', y='income', scatter_kws={'alpha': 0.5}, line_kws={'color': 'red'})

plt.xlabel('Age', **label_font)
plt.ylabel('Income', **label_font)
plt.title('Income vs Age with Trend Line', **title_font)
plt.show()

Findings:

• There is no significant trend between age and income.
• Income does not consistently increase with age.

Boxplot of income grouped by age group

• Boxplot of income across age groups

plt.figure(figsize=(12, 6))
sns.boxplot(data=df[df['income'] > 0], x='age_group', y='income', hue='age_group', palette='viridis', showfliers=False, legend=False)

plt.xlabel('Age Group', **label_font)
plt.ylabel('Income', **label_font)
plt.title('Income Distribution Across Age Groups', **title_font)
plt.show()

Findings:

• 36-45 age group has the highest median income.
• The presence of outliers in older groups suggests some users report significantly higher incomes.

Income variation by gender and status

• Barplot of income across gender and status

plt.figure(figsize=(12, 6))
sns.barplot(data=df[df['income'] > 0], x='status', y='income', hue='gender', palette='coolwarm')

plt.xlabel('Status', **label_font)
plt.ylabel('Income', **label_font)
plt.title('Income Variation by Gender and Status', **title_font)
plt.legend(title='Gender')
plt.show()

Findings:

• Single men report higher incomes than single women.
• Married users have a higher average income compared to other groups.

Pets and Preferences

• Pets are often a key lifestyle preference and compatibility factor.
• Analyzing pet preferences across demographics can provide insights for filters or recommendations.

To create a bar chart showing the distribution of pets categories (e.g., likes dogs, likes cats) based on preferences that are most common

import matplotlib.pyplot as plt
import seaborn as sns

pet_distribution = df['pets'].value_counts()

plt.figure(figsize=(10, 6))
pet_distribution = df['pets'].value_counts()
sns.barplot(x=pet_distribution.index, y=pet_distribution.values, hue=pet_distribution.index, 
            palette='pastel', dodge=False, legend=False, edgecolor='black')
plt.xlabel('Pet Categories', fontsize=12)
plt.ylabel('Count', fontsize=12)
plt.title('Distribution of Pet Categories', fontsize=14)
plt.xticks(rotation=45)
plt.show()

Findings:

• The most common pet preference is "likes dogs and likes cats".
• Very few users dislike pets.

How do pet preferences vary across gender and age_group?

pet_distribution_by_group = df.groupby(['age_group', 'gender', 'pets']).size().reset_index(name='count')

plt.figure(figsize=(12, 6))
sns.barplot(data=pet_distribution_by_group, x="age_group", y="count", hue="pets", 
            palette="muted", dodge=True, errorbar=None)  # `ci=None` changed to `errorbar=None`

plt.xlabel("Age Group", fontsize=12)
plt.ylabel("Number of Users", fontsize=12)
plt.title("Pet Preferences by Age Group and Gender", fontsize=14)
plt.legend(title="Pet Preference", fontsize=10)
plt.xticks(rotation=45)
plt.show()

C:\Program Files\KMSpico\temp\ipykernel_10508\2670204561.py:1: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behavior or observed=True to adopt the future default and silence this warning.
  pet_distribution_by_group = df.groupby(['age_group', 'gender', 'pets']).size().reset_index(name='count')

Findings:

• Users in the 19-25, 26-35, and 46+ groups mostly dislike pets.
• Gender distribution remains relatively balanced.

Signs and Personality

To create a chart showing the distribution of zodiac signs (sign)

• Used bar chart instead of pie chart (better for large categories)

sign_distribution = df['sign'].value_counts(normalize=True) * 100

plt.figure(figsize=(12, 8))
sns.barplot(x=sign_distribution.index, y=sign_distribution.values, palette="viridis")
plt.xlabel('Zodiac Signs', fontsize=12)
plt.ylabel('Percentage (%)', fontsize=12)
plt.title('Distribution of Zodiac Signs', fontsize=14)
plt.xticks(rotation=90)
plt.show()

C:\Program Files\KMSpico\temp\ipykernel_10508\1950768638.py:4: FutureWarning: 

Passing `palette` without assigning `hue` is deprecated and will be removed in v0.14.0. Assign the `x` variable to `hue` and set `legend=False` for the same effect.

  sns.barplot(x=sign_distribution.index, y=sign_distribution.values, palette="viridis")

Findings:

• The zodiac signs are almost equally distributed.
• A bar chart provides better visibility compared to a pie chart.

How does sign vary across gender?

plt.figure(figsize=(14, 8))
sign_across_gender = df.groupby(['sign', 'gender']).size().reset_index(name='count')
sns.barplot(data=sign_across_gender, x='sign', y='count', hue='gender', palette='coolwarm')

plt.title('Distribution of Zodiac Signs across Gender', fontsize=14)
plt.xlabel('Zodiac Sign', fontsize=12)
plt.ylabel('Number of Users', fontsize=12)
plt.xticks(rotation=90)
plt.legend(title='Gender', fontsize=10)
plt.show()

Findings:

• No significant difference in zodiac sign distribution between genders.

How does sign vary across status?

plt.figure(figsize=(14, 8))
sign_across_status = df.groupby(['sign', 'status']).size().reset_index(name='count')
sns.barplot(data=sign_across_status, x='sign', y='count', hue='status', palette='crest')

plt.title('Distribution of Zodiac Signs across Status', fontsize=14)
plt.xlabel('Zodiac Sign', fontsize=12)
plt.ylabel('Number of Users', fontsize=12)
plt.xticks(rotation=90)
plt.legend(title='Status', fontsize=10)
plt.grid(True, axis='y', linestyle='dashed', alpha=0.5)
plt.show()

Findings:

• Zodiac sign distribution across relationship status does not show any major patterns.