10 Prompts for Data Science: Analysis, Visualization, and Pandas

I have a CSV file named 'sales_data.csv' with columns: date, product_id, revenue, quantity, region. Write Pandas code to:
- Load the file with correct data types
- Display the first 5 rows
- Show summary statistics for numeric columns
- Check for missing values
- Print the number of unique values per column

import pandas as pd

df = pd.read_csv('sales_data.csv', parse_dates=['date'])
print(df.head())
print(df.describe())
print(df.isnull().sum())
print(df.nunique())

Given a DataFrame 'df' with columns: 'age', 'salary', 'email', 'signup_date'. Some 'age' values are negative, 'salary' has outliers above 1,000,000, and 'email' contains duplicates. Write Pandas code to:
- Replace negative ages with NaN
- Cap salary at the 99th percentile
- Drop duplicate emails keeping the first occurrence
- Convert signup_date to datetime and fill missing dates with the median date

df['age'] = df['age'].apply(lambda x: x if x >= 0 else None)
cap = df['salary'].quantile(0.99)
df['salary'] = df['salary'].clip(upper=cap)
df = df.drop_duplicates(subset='email', keep='first')
df['signup_date'] = pd.to_datetime(df['signup_date'], errors='coerce')
median_date = df['signup_date'].median()
df['signup_date'] = df['signup_date'].fillna(median_date)

Using the DataFrame 'df' with columns 'region', 'product', 'revenue', 'quantity', create:
- A grouped aggregation showing total revenue and average quantity per region
- A pivot table with regions as rows, products as columns, and sum of revenue as values
- A second pivot table with the same structure but showing count of transactions

grouped = df.groupby('region').agg({'revenue': 'sum', 'quantity': 'mean'})
pivot_rev = df.pivot_table(index='region', columns='product', values='revenue', aggfunc='sum', fill_value=0)
pivot_cnt = df.pivot_table(index='region', columns='product', values='revenue', aggfunc='count', fill_value=0)

The DataFrame 'df' has a datetime index and a column 'sales'. Write code to:
- Resample the data to weekly frequency using the mean
- Plot the original daily sales and the weekly resampled series on the same figure
- Add a rolling 4-week average to the weekly plot
- Highlight the period with maximum weekly sales

import matplotlib.pyplot as plt

weekly = df['sales'].resample('W').mean()
rolling = weekly.rolling(window=4).mean()

plt.figure(figsize=(12,6))
plt.plot(df.index, df['sales'], alpha=0.5, label='Daily')
plt.plot(weekly.index, weekly, label='Weekly Mean')
plt.plot(weekly.index, rolling, label='4-Week Rolling Avg')
plt.axvline(x=weekly.idxmax(), color='red', linestyle='--', label='Max Week')
plt.legend()
plt.show()

I have two groups in my DataFrame: 'control' and 'treatment' in column 'group', and a metric 'conversion_rate'. Write Python code to:
- Perform a two-sample t-test (assuming unequal variance) using scipy.stats
- Calculate Cohen's d effect size
- Create a boxplot comparing the two groups using Seaborn
- Print a summary table with mean, std, and sample size for each group

from scipy import stats
import seaborn as sns
import numpy as np

ctrl = df[df['group']=='control']['conversion_rate']
treat = df[df['group']=='treatment']['conversion_rate']

t_stat, p_val = stats.ttest_ind(ctrl, treat, equal_var=False)
cohens_d = (treat.mean() - ctrl.mean()) / np.sqrt((ctrl.var() + treat.var()) / 2)

sns.boxplot(x='group', y='conversion_rate', data=df)
plt.show()

summary = df.groupby('group')['conversion_rate'].agg(['mean', 'std', 'count'])
print(summary)

Create a publication-ready line plot using Matplotlib with the following specifications:
- Data: x = [1,2,3,4,5], y = [2,4,6,8,10]
- Use a clean sans-serif font (e.g., 'Arial')
- Set figure size to 8x5 inches
- Add grid lines with alpha=0.3
- Label axes with units in brackets: 'Time (s)' and 'Velocity (m/s)'
- Add a legend in the upper left corner
- Save the figure as a high-resolution PNG (300 dpi)

import matplotlib.pyplot as plt

plt.rcParams['font.family'] = 'Arial'
plt.figure(figsize=(8,5))
plt.plot([1,2,3,4,5], [2,4,6,8,10], label='Measured')
plt.xlabel('Time (s)')
plt.ylabel('Velocity (m/s)')
plt.grid(alpha=0.3)
plt.legend(loc='upper left')
plt.savefig('plot.png', dpi=300)
plt.show()

Using the Iris dataset (seaborn.load_dataset('iris')), create a pairplot that:
- Colors points by species
- Uses a custom color palette: ['#FF6B6B', '#4ECDC4', '#45B7D1']
- Adds a kernel density estimate on the diagonal
- Changes marker style for each species (e.g., 'o', '^', 's')
- Saves the figure as 'iris_pairplot.png'

import seaborn as sns

iris = sns.load_dataset('iris')
palette = ['#FF6B6B', '#4ECDC4', '#45B7D1']
markers = {'setosa': 'o', 'versicolor': '^', 'virginica': 's'}
sns.pairplot(iris, hue='species', palette=palette, diag_kind='kde', markers=markers)
plt.savefig('iris_pairplot.png')
plt.show()

I have a 5GB CSV file 'big_data.csv'. Write code to:
- Read the file in chunks of 100,000 rows using Pandas' read_csv with chunksize
- Calculate the mean of column 'value' across all chunks
- Also compute the 95th percentile of 'value' using incremental statistics (without loading all data)
- Print progress every 10 chunks

chunksize = 100000
sum_val = 0
count = 0
percentile_data = []

for i, chunk in enumerate(pd.read_csv('big_data.csv', chunksize=chunksize)):
    sum_val += chunk['value'].sum()
    count += len(chunk)
    percentile_data.append(chunk['value'].quantile(0.95))
    if i % 10 == 0:
        print(f'Processed chunk {i}')

mean_val = sum_val / count
overall_percentile = np.mean(percentile_data)  # approximate
print(f'Mean: {mean_val}, Approx 95th Percentile: {overall_percentile}')

Write Python code to generate an HTML report from a DataFrame 'df' with columns 'month', 'revenue', 'cost'. The report should:
- Include a table showing month, revenue, cost, profit (revenue - cost)
- Highlight rows where profit is negative in red
- Add a bar chart of revenue and cost using Matplotlib, embedded as a base64 PNG
- Use a Jinja2 template with CSS styling
- Save the final report as 'report.html'

from jinja2 import Template
import base64
from io import BytesIO

# Calculate profit
df['profit'] = df['revenue'] - df['cost']

# Create plot
plt.figure()
df.plot(x='month', y=['revenue','cost'], kind='bar')
buf = BytesIO()
plt.savefig(buf, format='png')
buf.seek(0)
img_str = base64.b64encode(buf.read()).decode()

# Template
template_str = '''
<html>
<head><style>table {border-collapse: collapse;} td, th {border: 1px solid black; padding: 5px;}</style></head>
<body>
<h1>Monthly Report</h1>
<table>
<tr><th>Month</th><th>Revenue</th><th>Cost</th><th>Profit</th></tr>
{% for row in rows %}
<tr {% if row.profit < 0 %}style="background-color: red;"{% endif %}>
    <td>{{ row.month }}</td><td>{{ row.revenue }}</td><td>{{ row.cost }}</td><td>{{ row.profit }}</td>
</tr>
{% endfor %}
</table>
<img src="data:image/png;base64,{{ img }}" />
</body></html>
'''
template = Template(template_str)
html = template.render(rows=df.to_dict('records'), img=img_str)

with open('report.html', 'w') as f:
    f.write(html)

Write Python code to fetch daily exchange rates (USD to EUR) from the Frankfurter API (https://api.frankfurter.app) for the last 30 days. Then:
- Load the JSON response into a Pandas DataFrame
- Calculate the 7-day moving average
- Plot the original rates and moving average
- Print the date with the highest rate

import requests
import pandas as pd
import matplotlib.pyplot as plt

url = 'https://api.frankfurter.app/2026-06-13..2026-07-13?from=USD&to=EUR'
response = requests.get(url)
data = response.json()
rates = data['rates']
df = pd.DataFrame(list(rates.items()), columns=['date', 'rate'])
df['date'] = pd.to_datetime(df['date'])
df = df.sort_values('date')
df['ma7'] = df['rate'].rolling(7).mean()

plt.plot(df['date'], df['rate'], label='Original')
plt.plot(df['date'], df['ma7'], label='7-day MA')
plt.legend()
plt.show()

max_date = df.loc[df['rate'].idxmax(), 'date']
print(f'Highest rate on: {max_date.date()}')