[Kaggle] Screen Time vs Mental Wellness Survey

Author

SEOYEON CHOI

Published

September 22, 2025

Reference

data

Import

import pandas as pd
import plotly.express as px
from scipy.stats import spearmanr, pearsonr

Data

df = pd.read_csv('../../../delete/ScreenTime vs MentalWellness.csv').iloc[:, :-1]
df
user_id age gender occupation work_mode screen_time_hours work_screen_hours leisure_screen_hours sleep_hours sleep_quality_1_5 stress_level_0_10 productivity_0_100 exercise_minutes_per_week social_hours_per_week mental_wellness_index_0_100
0 U0001 33 Female Employed Remote 10.79 5.44 5.35 6.63 1 9.3 44.7 127 0.7 9.3
1 U0002 28 Female Employed In-person 7.40 0.37 7.03 8.05 3 5.7 78.0 74 2.1 56.2
2 U0003 35 Female Employed Hybrid 9.78 1.09 8.69 6.48 1 9.1 51.8 67 8.0 3.6
3 U0004 42 Male Employed Hybrid 11.13 0.56 10.57 6.89 1 10.0 37.0 0 5.7 0.0
4 U0005 28 Male Student Remote 13.22 4.09 9.13 5.79 1 10.0 38.5 143 10.1 0.0
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
395 U0396 26 Female Student Remote 6.43 2.99 3.44 7.75 1 5.9 64.6 252 7.6 39.3
396 U0397 16 Male Self-employed Remote 9.59 5.44 4.15 5.57 1 10.0 47.4 99 7.0 3.5
397 U0398 40 Male Student Remote 8.72 2.36 6.36 7.56 1 9.4 57.3 193 10.1 6.6
398 U0399 29 Female Retired Hybrid 5.04 0.94 4.10 7.32 1 7.1 63.6 97 12.1 21.0
399 U0400 39 Female Employed Remote 8.86 3.82 5.04 6.55 1 10.0 41.4 88 9.8 8.4

400 rows × 15 columns

  • screen_time_hours은 work_screen_hours와 leisure_screen_hours의 합임을 확인
(df["screen_time_hours"] == df["work_screen_hours"] + df["leisure_screen_hours"]).all()
False
diff[diff>0]
Series([], dtype: float64)
diff = round(df["screen_time_hours"] - (df["work_screen_hours"] + df["leisure_screen_hours"]),2)
df_diff = df[diff != 0]
df_diff
user_id age gender occupation work_mode screen_time_hours work_screen_hours leisure_screen_hours sleep_hours sleep_quality_1_5 stress_level_0_10 productivity_0_100 exercise_minutes_per_week social_hours_per_week mental_wellness_index_0_100 
df.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 400 entries, 0 to 399
Data columns (total 15 columns):
 #   Column                       Non-Null Count  Dtype  
---  ------                       --------------  -----  
 0   user_id                      400 non-null    object 
 1   age                          400 non-null    int64  
 2   gender                       400 non-null    object 
 3   occupation                   400 non-null    object 
 4   work_mode                    400 non-null    object 
 5   screen_time_hours            400 non-null    float64
 6   work_screen_hours            400 non-null    float64
 7   leisure_screen_hours         400 non-null    float64
 8   sleep_hours                  400 non-null    float64
 9   sleep_quality_1_5            400 non-null    int64  
 10  stress_level_0_10            400 non-null    float64
 11  productivity_0_100           400 non-null    float64
 12  exercise_minutes_per_week    400 non-null    int64  
 13  social_hours_per_week        400 non-null    float64
 14  mental_wellness_index_0_100  400 non-null    float64
dtypes: float64(8), int64(3), object(4)
memory usage: 47.0+ KB
  • 문자형 컬럼
char_cols = df.select_dtypes(include=['object', 'string']).columns
for col in char_cols:
    print(f"\n--- {col} (categorical) ---")
    print(df[col].value_counts())

--- user_id (categorical) ---
U0001    1
U0264    1
U0274    1
U0273    1
U0272    1
        ..
U0131    1
U0130    1
U0129    1
U0128    1
U0400    1
Name: user_id, Length: 400, dtype: int64

--- gender (categorical) ---
Female              222
Male                170
Non-binary/Other      8
Name: gender, dtype: int64

--- occupation (categorical) ---
Employed         207
Student          107
Self-employed     45
Unemployed        27
Retired           14
Name: occupation, dtype: int64

--- work_mode (categorical) ---
Remote       150
Hybrid       146
In-person    104
Name: work_mode, dtype: int64
  • 숫자형 컬럼
num_cols = df.select_dtypes(include=['number']).columns
print("\n--- Numeric summary ---")
print(df[num_cols].describe().T)

--- Numeric summary ---
                             count        mean        std    min      25%  \
age                          400.0   29.777500   7.466080  16.00  24.0000   
screen_time_hours            400.0    9.024900   2.491058   1.00   7.3725   
work_screen_hours            400.0    2.183075   1.931321   0.11   0.6950   
leisure_screen_hours         400.0    6.841825   2.220896   0.89   5.4600   
sleep_hours                  400.0    7.013175   0.852421   4.64   6.3975   
sleep_quality_1_5            400.0    1.397500   0.652348   1.00   1.0000   
stress_level_0_10            400.0    8.150500   2.094844   0.00   6.9000   
productivity_0_100           400.0   54.306500  15.020054  20.60  43.6000   
exercise_minutes_per_week    400.0  109.810000  70.007045   0.00  58.0000   
social_hours_per_week        400.0    7.905000   4.909632   0.00   4.5750   
mental_wellness_index_0_100  400.0   20.326750  20.376793   0.00   3.6750   

                                 50%       75%     max  
age                           30.000   35.0000   60.00  
screen_time_hours              9.090   10.4950   19.17  
work_screen_hours              1.455    3.0125   12.04  
leisure_screen_hours           6.700    8.4175   13.35  
sleep_hours                    7.030    7.6400    9.74  
sleep_quality_1_5              1.000    2.0000    4.00  
stress_level_0_10              8.800   10.0000   10.00  
productivity_0_100            51.750   63.0000  100.00  
exercise_minutes_per_week    103.000  157.0000  372.00  
social_hours_per_week          7.750   11.0250   23.90  
mental_wellness_index_0_100   14.800   30.6500   97.00

Visualization

Character col check

fig = px.bar(
    df['gender'].value_counts().reset_index(),
    x='index', y='gender',
    labels={'index':'Gender', 'gender':'Count'},
    title='Gender Distribution',
    height=500, width= 700)
fig.show()
fig = px.bar(
    df['occupation'].value_counts().reset_index(),
    x='index', y='occupation',
    labels={'index':'occupation', 'occupation':'Count'},
    title='Occupation Distribution',
    height=500, width= 700)
fig.show()
fig = px.bar(
    df['work_mode'].value_counts().reset_index(),
    x='index', y='work_mode',
    labels={'index':'work_mode', 'work_mode':'Count'},
    title='work_mode Distribution',
    height=500, width= 700)
fig.show()

Gender and Work mode

  • Average Screen Time, work screen hours and leisure screen hours by Gender

성별 차이는 없어 보임

df_grouped = df.groupby("gender")[["screen_time_hours", "work_screen_hours","leisure_screen_hours"]].mean().reset_index()

fig = px.bar(
    df_grouped,
    x="gender", 
    y=["screen_time_hours", "work_screen_hours","leisure_screen_hours"],
    barmode="group",
    title="Average Screen Time by Gender",
    labels={"value": "Hours", "gender": "Gender", "variable": "Type"},
    width=700, height=500
)
fig.show()
  • Average Screen Time, work screen hours and leisure screen hours by work mode

remote로 하는 사람들이 당연히 work_xcreen_hours 가장 높았고 이에따라 screen time hours도 높았음

df_grouped = df.groupby("work_mode")[["screen_time_hours", "work_screen_hours","leisure_screen_hours"]].mean().reset_index()

fig = px.bar(
    df_grouped,
    x="work_mode", 
    y=["screen_time_hours", "work_screen_hours","leisure_screen_hours"],
    barmode="group",
    title="Average Screen Time by work_mode",
    labels={"value": "Hours", "work_mode": "work_mode", "variable": "Type"},
    width=700, height=500
)
fig.show()
  • sleep quality

다 왜 이렇게 낮아..ㅎㅎㅎ

fig = px.box(
    df,
    x="work_mode",
    y="sleep_quality_1_5",
    color="gender",  
    title="Stress Level by Work Mode and Gender",
    height=500,
    width=700
)
fig.show()
fig = px.histogram(
    df,
    x="sleep_quality_1_5",
    color="gender",  
    barmode="group",
    facet_col="work_mode", 
    title="Distribution of Sleep Quality by Gender and Work Mode",
    height=600,
    width=900
)
fig.show()
  • stress level

remote에 비해 in-person,hybrid 의 stress level이 낮네?

fig = px.box(
    df,
    x="work_mode",
    y="stress_level_0_10",
    color="gender",  
    title="Stress Level by Work Mode and Gender",
    height=500,
    width=700
)
fig.show()
  • productivity

in-person, hybrid가 remote에 비해 높은 경향

fig = px.box(
    df,
    x="work_mode",
    y="productivity_0_100",
    color="gender",   
    title="Productivity by Work Mode and Gender",
    height=500,
    width=700
)
fig.show()
  • excercise minutes per week

remote라고 많거나 하지 않네

fig = px.box(
    df,
    x="work_mode",
    y="exercise_minutes_per_week",
    color="gender",   
    title="Excercise minutes per week by Work Mode and Gender",
    height=500,
    width=700
)
fig.show()
  • Social hours per week

remote가 오히려 낮은 경향, hybrid가 높은 경향

fig = px.box(
    df,
    x="work_mode",
    y="social_hours_per_week",
    color="gender",   
    title="Social hours per week by Work Mode and Gender",
    height=500,
    width=700
)
fig.show()
  • Mental Wellness index

remote가 낮은경향, 그렇다고 나머지 높진 않음.

fig = px.box(
    df,
    x="work_mode",
    y="mental_wellness_index_0_100",
    color="gender",   
    title="Mental wellness index by Work Mode and Gender",
    height=500,
    width=700
)
fig.show()

Occupation

student들이 screen 시간이 길지만 leisure hours는 직장인들과 비슷하다

df_long = df.melt(
    id_vars="occupation", 
    value_vars=["screen_time_hours", "work_screen_hours", "leisure_screen_hours"],
    var_name="screen_type",
    value_name="hours"
)

fig = px.box(
    df_long,
    x="occupation",
    y="hours",
    color="screen_type",  # work, leisure, total 구분
    title="Screen Hours by Occupation",
    height=500,
    width=1000
)

fig.show()
  • Sleep Hours

sleep hours는 비슷

fig = px.box(
    df,
    x="occupation",
    y="sleep_hours",
    title="Sleep Hours by occupation",
    height=500, width= 700)
fig.show()
  • sleep quality

다 낮다..

fig = px.histogram(
    df,
    x="sleep_quality_1_5",
    color="occupation",  
    barmode="group",
    title="Distribution of Sleep Quality",
    height=500,
    width=800
)
fig.show()
  • Stress Level

Stress Level은 student가 평균적으로 높다

fig = px.box(
    df,
    x="occupation",
    y="stress_level_0_10",
    title="Stress Level by occupation",
    height=500, width= 700)
fig.show()
  • Productivity

Productivity는 학생이 제일 낮고 retired가 가장 높다

fig = px.box(
    df,
    x="occupation",
    y="productivity_0_100",
    title="Productivity by occupation",
    height=500, width= 700)
fig.show()
  • Excercise Minutes

Excercise Minutes는 unemployed가 평균적으로 가장 높고 그다음 retired, 다음은 비슷

fig = px.box(
    df,
    x="occupation",
    y="exercise_minutes_per_week",
    title="Excercise Minutes by occupation",
    height=500, width= 700)
fig.show()
  • Social Hours per week

다 비슷하고 student가 낮은 경향

fig = px.box(
    df,
    x="occupation",
    y="social_hours_per_week",
    title="Social Hours per week by occupation",
    height=500, width= 700)
fig.show()
  • Mental Wellness

Mental Wellness는 student가 평균적으로 가장 낮고, employed, self-employed가 그다음 낮고 retired, unemployed가 높은 그나저나 student 분산이 꽤나 높을듯

fig = px.box(
    df,
    x="occupation",
    y="mental_wellness_index_0_100",
    title="Mental Wellness by occupation",
    height=500, width= 700)
fig.show()

Trend Check

전반적 경향: screen time이 늘어날수록 sleep 이 줄어든다.

fig = px.scatter(
    df, 
    x="screen_time_hours", 
    y="sleep_hours", 
    trendline="ols", 
    # color='gender',
    title="Screen Time vs Sleep Hours",
    height=500, width= 700)
fig.show()

stress level 이 높을수록 screen time시간이 길고, sleep hours는 줄어드는 trend

fig = px.scatter(
    df, 
    x="screen_time_hours", 
    y="sleep_hours", 
    trendline="ols", 
    color='stress_level_0_10',
    title="Screen Time vs Sleep Hours",
    height=500, width=700,
    labels={"stress_level_0_10": "Stress Level (0-10)"}
)
fig.show()

productivity가 낮을수록 screen time hour도 갈고 sleep hour는 낮아지는 경향

fig = px.scatter(
    df, 
    x="screen_time_hours", 
    y="sleep_hours", 
    trendline="ols", 
    color='productivity_0_100',
    title="Screen Time vs Sleep Hours",
    height=500, width= 700,
    labels={"productivity_0_100": "Productivity (0-100)"}
)
fig.show()
  • Mental Wellness

Mental Wellness가 낮을수록 screen time도 길고 sleep hour도 낮아지는 경향

fig = px.scatter(
    df, 
    x="screen_time_hours", 
    y="sleep_hours", 
    trendline="ols", 
    color='mental_wellness_index_0_100',
    title="Screen Time vs Sleep Hours",
    height=500, width= 900,
    labels={"mental_wellness_index_0_100": "Mental Wellness (0-100)"}
)
fig.show()
  • screen_time_hours, sleep_hours에 따라 graph 확인해본 결과 stress_level_0_10, productivity_0_100, mental_wellness_index_0_100가 영향이 있어보였다.

Correlation

corr = df[['screen_time_hours','work_screen_hours','leisure_screen_hours','sleep_hours',
           'sleep_quality_1_5','stress_level_0_10','productivity_0_100',
           'exercise_minutes_per_week','social_hours_per_week','mental_wellness_index_0_100']].corr()
px.imshow(corr, text_auto=True, title="Correlation Heatmap",
    height=1000, width= 1000)
  • 1부터 5점 척도로 이루어진 sleep_quality_1_5 때문에 spearman correlation으로
spearman_corr, spearman_p = spearmanr(df['sleep_hours'], df['sleep_quality_1_5'])
print(f"Spearman correlation: {spearman_corr:.3f}, p-value: {spearman_p:.3f}")
Spearman correlation: 0.626, p-value: 0.000
  • 잠자는 시간과 수면의 질의 상관계수는 0.626, p-value<0.000으로 연관 있다는 결과,
  • 아래 그림으로 봐도 양의 상관관계가 있는 것으로 보인다.
fig = px.box(
    df, 
    x='sleep_quality_1_5', 
    y='sleep_hours',
    points="all",  
    title="Sleep Hours by Sleep Quality (1-5)",
    height=500, width= 900
)
fig.show()
  • stress level vs mental wellness
spearman_corr, spearman_p = spearmanr(df['mental_wellness_index_0_100'], df['stress_level_0_10'])
print(f"Spearman correlation: {spearman_corr:.3f}, p-value: {spearman_p:.3f}")
Spearman correlation: -0.871, p-value: 0.000
  • 강한 음의 상관관계, 유의미함
px.scatter(df, x="mental_wellness_index_0_100", y="stress_level_0_10", trendline="ols", height=500, width= 900)
  • mental wellness vs productivity
spearman_corr, spearman_p = spearmanr(df['mental_wellness_index_0_100'], df['productivity_0_100'])
print(f"Spearman correlation: {spearman_corr:.3f}, p-value: {spearman_p:.3f}")
Spearman correlation: 0.869, p-value: 0.000
  • 강한 양의 상관, 유의미함
px.scatter(df, x="mental_wellness_index_0_100", y="productivity_0_100", trendline="ols", height=500, width= 900)
  • productivity vs stress level
spearman_corr, spearman_p = spearmanr(df['productivity_0_100'], df['stress_level_0_10'])
print(f"Spearman correlation: {spearman_corr:.3f}, p-value: {spearman_p:.3f}")
Spearman correlation: -0.856, p-value: 0.000
  • 강한 음의 상관, 생산성이 커질수록 스트레스 수준이 감소
px.scatter(df, x="productivity_0_100", y="stress_level_0_10", height=500, width= 900)
  • screen time vs stress level
spearman_corr, spearman_p = spearmanr(df['screen_time_hours'], df['stress_level_0_10'])
print(f"Spearman correlation: {spearman_corr:.3f}, p-value: {spearman_p:.3f}")
Spearman correlation: 0.719, p-value: 0.000
  • 강한 양의 상관, 유의미함, 스크린 타임이 길수로 스트레스가 많아짐
px.scatter(df, x="screen_time_hours", y="stress_level_0_10", height=500, width= 900)