KEMBAR78
TSA Assignment | PDF
Scribd
Upload
15 views2 pages

TSA Assignment

The document outlines a Python script that analyzes monthly sales data for three business units (BU1, BU2, BU3) from a CSV file. It includes steps for loading the data, creating a datetime index, checking for stationarity using the ADF test, fitting SARIMAX models, and forecasting sales for the first three months of 2018. Finally, it visualizes both historical and forecasted sales data for each business unit.

Uploaded by

Finn Hourigan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
15 views2 pages

TSA Assignment

The document outlines a Python script that analyzes monthly sales data for three business units (BU1, BU2, BU3) from a CSV file. It includes steps for loading the data, creating a datetime index, checking for stationarity using the ADF test, fitting SARIMAX models, and forecasting sales for the first three months of 2018. Finally, it visualizes both historical and forecasted sales data for each business unit.

Uploaded by

Finn Hourigan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
You are on page 1/ 2

import pandas as pd

import numpy as np
import matplotlib.pyplot as plt
from arch.unitroot import ADF
from statsmodels.tsa.statespace.sarimax import SARIMAX

#2. Load data and create proper datetime column


df = pd.read_csv('USA FIRM SALES DATA.csv')

print(df.tail(5))

# Combine Year and Month into datetime


df['Date'] = pd.to_datetime(df['Year'].astype(str) + ' ' + df['Month'], format='%Y
%B')

# Set datetime as index


df.set_index('Date', inplace=True)
df.index.freq = 'MS'

# Drop original Year/Month


df = df[['BU1', 'BU2', 'BU3']]

# 3. Create Time Series and Plot


df.plot(title="Monthly Sales for BU1, BU2, BU3")
plt.ylabel("Sales (in million USD)")
plt.show()

# 4. Stationarity check
def adf_test(series, name):
result = ADF(series)
print(f"\nADF Test for {name}")
print("ADF Statistic:", result.stat)
print("p-value:", result.pvalue)
print("=> Stationary" if result.pvalue < 0.05 else "=> Non-Stationary")

# Run ADF test for each Business Unit


for bu in df.columns:
adf_test(df[bu], bu)

# 5. Fit SARIMAX models

# Create a dictionary to store fitted models


models = {}
predictions = {}

# Forecast horizon: Jan, Feb, Mar 2018


forecast_steps = 3

# Fit a SARIMAX model for each BU


for bu in df.columns:
model = SARIMAX(df[bu], order=(1, 1, 1), seasonal_order=(1,1,1,12),
enforce_stationarity=False, enforce_invertibility=False)
model_fit = model.fit(disp=False)
print(f'AIC : {model_fit.aic}')
models[bu] = model_fit

# Forecast next 3 months


pred = model_fit.get_forecast(steps=forecast_steps)
pred_mean = pred.predicted_mean
predictions[bu] = pred_mean

print(f"\nForecast for {bu} (Jan-Mar 2018):")


print(pred_mean)

# 6. Final forecast

forecast_df = pd.DataFrame(predictions)
forecast_df.index = pd.date_range(start='2018-01-01', periods=3, freq='MS')
print("\nAll BU Forecasts (Jan to Mar 2018):")
print(forecast_df)

# 7. Plot forcasts
import matplotlib.pyplot as plt

# Plot observed + forecast for each BU


for bu in df.columns:
plt.figure(figsize=(10, 4))

# Plot historical sales


df[bu].plot(label='Historical', color='blue')

# Plot forecasted sales


forecast_df[bu].plot(label='Forecast (Jan–Mar 2018)', color='orange',
linestyle='--')

plt.title(f"Sales Forecast for {bu}")


plt.xlabel("Date")
plt.ylabel("Sales (in million USD)")
plt.legend()
plt.grid(True)
plt.tight_layout()
plt.show()

You might also like