Topic 2 – Monthly Business Sales Time Series

Dataset

• Source: Business Sales Time Series Starter – OpenDataBay
• Link: https://www.opendatabay.com/data/ai-ml/f86d74d9-4656-4313-94d7-0cbd12bb7ffd

Download Instructions

1. Open the link above.
2. Click "Download CSV" (e.g. Month_Value_1.csv).
3. Save as data/business_sales.csv.

Data Loading

import pandas as pd
df = pd.read_csv("data/business_sales.csv")
df["Month"] = pd.to_datetime(df["Month"])
df = df.set_index("Month").sort_index()

Implementation Steps

1. Data Exploration

• Load and inspect the dataset structure
• Identify the target variable (sales/revenue column)
• Check data range and frequency (monthly)
• Examine basic statistics and data quality

2. Exploratory Data Analysis (EDA)

• Plot monthly sales over time
• Identify trends (increasing, decreasing, stable)
• Detect seasonality patterns (monthly, quarterly, yearly)
• Perform time series decomposition (additive or multiplicative)
• Calculate ACF/PACF to understand autocorrelation

3. Stationarity Analysis

• Visual inspection (rolling mean and variance)
• Perform Augmented Dickey-Fuller (ADF) test
• Apply differencing if non-stationary
• Consider seasonal differencing if strong seasonality exists

4. Model Building

• Classical Methods:
• ARIMA models (identify p, d, q from ACF/PACF)
• SARIMA models (for seasonal patterns)
• Exponential Smoothing (Holt-Winters)
• Model Selection:
• Use AIC/BIC for comparison
• Cross-validation on validation set

5. Model Evaluation

• Split data: training (70%), validation (15%), test (15%)
• Generate forecasts and compare with actual values
• Calculate error metrics: MAE, RMSE, MAPE
• Visualize forecasts vs actuals
• Check residual diagnostics

6. Forecasting

• Generate future monthly forecasts
• Include prediction intervals
• Visualize with historical data

Expected Deliverables

1. EDA Report:

• Time series plots with trend/seasonality
• Decomposition plots
• ACF/PACF analysis
• Stationarity test results

1. Model Results:

• Selected model with parameters
• Model diagnostics (residual plots, ACF of residuals)
• Performance metrics table
• Forecast plots

1. Code:

• Complete Python notebook with all steps
• Reusable functions for preprocessing and modeling

Tips

• Monthly data often shows strong seasonality (yearly patterns)
• Consider multiplicative models if variance increases with level
• Use SARIMA for seasonal patterns (e.g., SARIMA(1,1,1)(1,1,1)12)
• Compare multiple models (ARIMA, SARIMA, Exponential Smoothing)
• Business sales may have external factors (holidays, promotions) - note these in analysis