Project 2: Sales Forecasting for Retail

Overview

This project involved developing a time series forecasting model to predict future sales for a retail company. The objective was to optimize inventory management and improve sales strategies.

Key Features

• Data Exploration: Analyzing sales trends and seasonality.
• Model Development: Using ARIMA and Prophet models for forecasting.
• Model Evaluation: Comparing model performance using RMSE and MAPE.

R Code Snippet

# Load necessary libraries
library(forecast)
library(prophet)

# Simulate retail sales data
set.seed(123)
n_months <- 72  # 6 years of monthly data
dates <- seq.Date(from = as.Date("2015-01-01"), by = "month", length.out = n_months)
sales <- rpois(n_months, lambda = 200) + seq(100, 300, length.out = n_months)

# Create data frame
sales_data <- data.frame(Date = dates, Sales = sales)

# Data preprocessing
sales_ts <- ts(sales_data$Sales, start = c(2015, 1), frequency = 12)

# ARIMA model
arima_model <- auto.arima(sales_ts)
arima_forecast <- forecast(arima_model, h = 12)

# Prophet model
df <- data.frame(ds = sales_data$Date, y = sales_data$Sales)
prophet_model <- prophet(df)
future <- make_future_dataframe(prophet_model, periods = 12, freq = "month")
prophet_forecast <- predict(prophet_model, future)

# Plot forecasts
plot(arima_forecast)

prophet_plot_components(prophet_model, prophet_forecast)

```