Python for Data Analysis: From Raw CSV to Business Insight

Introduction

A CSV file containing 500,000 rows can crash standard spreadsheet software. For data analysts, Python provides a memory-efficient, highly automated environment to clean tables, calculate complex customer segment metrics, and export data in seconds. Today we will walk through a real-world Python workflow using Pandas and NumPy to turn raw data into insights.

Setting Up the Pandas Environment

First, we import the core data libraries and read the transaction log dataset. Pandas reads datasets as a “DataFrame” (a two-dimensional table representation):

        SQL / PYTHON QUERY
        
          
          Copy
        
      
import pandas as pd
import numpy as np

# Load raw transaction csv
df = pd.read_csv("sales_data_raw.csv")
print(df.head())

1. Cleaning Messy Datasets

Messy files often contain duplicate transactions and empty fields. We clean them programmatically:

        SQL / PYTHON QUERY
        
          
          Copy
        
      
# Remove duplicate rows
df.drop_duplicates(inplace=True)

# Fill missing transaction values with column median
df['amount'] = df['amount'].fillna(df['amount'].median())

# Standardize date types
df['order_date'] = pd.to_datetime(df['order_date'])

2. Cohort Feature Engineering

Next, we create new metrics. For instance, calculating purchase values above ₹10,000 as high-value tags:

        SQL / PYTHON QUERY
        
          
          Copy
        
      
df['is_high_value'] = np.where(df['amount'] >= 10000, 'Yes', 'No')

3. Summarizing the Data (GROUP BY)

We segment performance metrics by acquisition channel to see which advertising channel returns the highest customer spending:

        SQL / PYTHON QUERY
        
          
          Copy
        
      
summary = df.groupby('acquisition_channel').agg(
    total_sales=('amount', 'sum'),
    average_order=('amount', 'mean'),
    customer_count=('customer_id', 'nunique')
).reset_index()

print(summary)

4. Extracting Business Insights

We can sort the results to identify key business drivers. Here, we find the highest performing channels:

        SQL / PYTHON QUERY
        
          
          Copy
        
      
# Sort summary by total sales
sorted_summary = summary.sort_values(by='total_sales', ascending=False)
print(sorted_summary)

5. Plotting a Visual Trend

Using Matplotlib, we draw a quick line graph tracking monthly revenue trajectory to spot seasonal drops:

        SQL / PYTHON QUERY
        
          
          Copy
        
      
import matplotlib.pyplot as plt

monthly_sales = df.groupby(df['order_date'].dt.to_period('M'))['amount'].sum()
monthly_sales.plot(kind='line', marker='o', color='#0ea5e9')
plt.title("Monthly Revenue Performance")
plt.xlabel("Month")
plt.ylabel("Revenue (₹)")
plt.savefig("monthly_sales_trend.png")

Why Learn Python Over Standard Spreadsheets?

Unlike Excel, Python files are repeatable recipes. If you receive a new monthly sales CSV, you can simply run your script again to clean, group, analyze, and chart the data in milliseconds. This reproducibility is what saves hours in professional business intelligence workflows.

Next Steps

In our SBS Analytics Lab, we teach Pandas, NumPy, and Matplotlib by coding real business pipelines and automating data transformations.