Part1 : Provided a dataset of volume sales of products from 2019 to 2022 run an extensive exploratory data analysis including the following: . 1. Data Quality & Structure Checks Missing values, duplicates, negative sales, outliers Date consistency (no gaps, proper frequency, handling holidays/weekends) 2. Descriptive Statistics Overall distribution of daily sales (mean, median, std, skewness, kurtosis) By dimension: product, customer Identify top products per customer by volume 3. Time Series Exploration Trend: long-term upward or downward movement Seasonality: daily/weekly patterns (weekdays vs weekends), monthly, quarterly, yearly cycles Rolling averages (7-day, 30-day) to smooth patterns 4. Visualization Layer Time series plots: raw daily sales, moving averages Boxplots: distribution of sales by weekday or month Histograms/density plots: sales distribution 5. Anomaly & Outlier Detection Unusual spikes/drops Use Z-scores or interquartile ranges to flag anomalies 6. Correlation & Drivers of Sales Correlation if needed 7. Performance Metrics (Baseline) Set benchmarks to prepare for forecasting models: Average daily sales per SKU/store Volatility (Coefficient of Variation) Baseline forecast error (e.g., naïve forecast MAPE)
EDA Deliverables : By the end of an extensive EDA, I should have: Clear understanding of demand patterns, seasonality, and anomalies Insights into drivers of sales (internal like price/promo, external like weather/events) Segmentation of products into high/medium/low performers A baseline performance snapshot to compare forecasting models against.
Part 2 : After cleaning the data based on the above analysis, run a linear regression-based model to prepare a sales volume forecast at product & customer level for 2022 in python or/and pyspark. Measure the accuracy by introducing quality measures and explain why have you introduced these measures.
