# Running the following line will install the 'pandasai' library
!pip install pandasai

# Import the pandas library, commonly used for data manipulation and analysis
import pandas as pd

# Import the pyplot module from the matplotlib library, used for plotting and visualization
import matplotlib.pyplot as plt

# Import the Agent class from the pandasai module, which can interface with large language models (LLMs)
from pandasai import Agent

# Import the SmartDataframe class from the pandasai module, which adds enhanced data manipulation features
from pandasai import SmartDataframe

# Import the OpenAI class within the pandasai.llm.openai submodule, which allows interfacing with OpenAI's language models
from pandasai.llm.openai import OpenAI

# Read data from a CSV file available at the given URL and store it in a DataFrame called sales_df
# The read_csv function fetches the data and structures it into a format suitable for analysis
sales_df = pd.read_csv('https://infinitepy.s3.amazonaws.com/samples/sales_data.csv')

# 'head()' is a built-in method in pandas DataFrame that returns the first n rows.
# By default, it returns the first 5 rows. You can specify a different number as an argument like 'head(10)' for the first 10 rows.
sales_df.head()

# Importing the 'userdata' module from google.colab, which allows you to access user-specific data.
from google.colab import userdata

# Fetching the OpenAI API key from the user's data stored within Google Colab.
api_token = userdata.get('OPENAI_API_KEY')

# Creating an instance of the OpenAI class by passing the API token.
# This instance will be used to interact with OpenAI's language model (LLM).
llm = OpenAI(api_token=api_token)

# Wrapping an existing DataFrame (sales_df) into a SmartDataframe object.
# SmartDataframe likely adds additional capabilities or configurations to the standard DataFrame.
# We pass the 'llm' instance as part of the configuration, possibly to enable natural language capabilities or other features powered by the OpenAI model.
sales_smart_df = SmartDataframe(sales_df, config={"llm": llm})

# Drop duplicate rows from the DataFrame 'sales_df' in place.
# This means any duplicate rows will be removed directly from the 'sales_df' object,
# and no copy of the DataFrame will be created.
sales_df.drop_duplicates(inplace=True)

# Fill any missing (NaN) values in the DataFrame 'sales_df' with 0.
# The 'inplace=True' parameter means that the DataFrame will be modified directly,
# without creating a copy.
sales_df.fillna(0, inplace=True)
sales_df

sales_smart_df.chat("Remove duplicates and fill missing values")

# Use groupby() to group the data by the 'Region' column
# Then, chain .sum() to calculate the total sales for each region
total_sales_region = sales_df.groupby('Region')['Sales'].sum()

# Print the result which shows the total sales per region
total_sales_region

sales_smart_df.chat("Compute total sales per region")

# Group the sales data by 'Product' and sum the 'Sales' for each product.
top_products_sales = sales_df.groupby('Product')['Sales'].sum()

# Sort the summed sales in descending order to get the best selling products at the top.
top_products_sales_sorted = top_products_sales.sort_values(ascending=False)

# Select the top 5 products with the highest sales.
top_5_products_sales_sorted = top_products_sales_sorted.head(5)

# Print the result to see the names and sales of the top 5 products.
top_5_products_sales_sorted

sales_smart_df.chat("Find the top 5 products by sales")

# Drop duplicates and forward fill missing values
sales_df.drop_duplicates(inplace=True)
sales_df.fillna(method='ffill', inplace=True)

# Define the correct chronological order for months
month_order = ['January', 'February', 'March', 'April', 'May', 'June', 'July', 'August', 'September', 'October', 'November', 'December']

# Convert the 'Month' column to a categorical type with the specified order
sales_df['Month'] = pd.Categorical(sales_df['Month'], categories=month_order, ordered=True)

# Pivot table to calculate total sales per month for each region
monthly_sales = sales_df.pivot_table(index='Month', columns='Region', values='Sales', aggfunc='sum')

# Filter the DataFrame to include only data from January to June
filtered_monthly_sales_df = monthly_sales.loc['January':'June']

# Plotting the monthly sales trend for each region
fig, ax = plt.subplots(figsize=(12, 8))
filtered_monthly_sales_df.plot(kind='line', marker='o', ax=ax)

# Adding grid lines
ax.grid(True, which='both', linestyle='--', linewidth=0.5)

# Additional plot settings
plt.title('Monthly Sales Trend per Region (January to June)')
plt.xlabel('Month')
plt.ylabel('Sales')
plt.legend(title='Region')
plt.xticks(rotation=45)
plt.show()

sales_smart_df.chat("Calculates the sales trend on a monthly basis for each region.")

# Compute the sales share for each product
sales_share = sales_df.groupby('Product')['Sales'].sum() / sales_df['Sales'].sum() * 100

# Plotting the sales share as a pie chart
sales_share.plot(kind='pie', autopct='%1.1f%%', figsize=(8, 8))
plt.title('Sales Share by Product')
plt.ylabel('')  # Hide the y-label
plt.show()

sales_smart_df.chat("Calculates the Sales Share by Product and create a pie chart")

🔔 Subscribe to the InfinitePy Newsletter for more resources and a step-by-step approach to learning Python, and stay up to date with the latest trends and practical tips.