Revenue Forecasting

Forecast revenue for products in the food and beverages industry using panel data with 50 products tracked over 100 time periods.

Dataset Exploration

The Food & Beverages dataset contains sales and inventory data for retail food products. Features include pricing, advertising spend, store traffic, competitor activity, and more. The target variable predicts revenue to optimize inventory management and pricing strategies.

Panel data structure: 50 products tracked over 100 time periods (5,000 total rows, 13 columns). Notice how Product 1 appears across multiple consecutive time periods — this is panel data.

Fit Data (first 5 rows)

Product	Time	Price	Add_spend	Store_traffic	Log_inventory	Competitor_price	Comp._promo	Price_X_Promo	Add_X_Price	Store_size	Cust_demo	Revenue
1	43831	3.37	465.27	2.56	4.12	3.78	0	0	1758.72	2290	3	145623
1	43832	3.73	423.33	3.06	3.64	3.61	1	3.73	1528.22	2290	3	142171.2
1	43833	3.39	555.51	2.86	4.02	2.94	0	0	1633.2	2290	3	145126.3
1	43834	3.41	487.45	2.79	4.07	3.58	0	0	1745.07	2290	3	145192.8
1	43835	3.12	525.51	3.17	3.97	3.63	0	0	1907.6	2290	3	147483.7

Forecast Data (first 3 rows)

The forecast file has the same structure but covers future periods without the Revenue column. DATFID uses this to know which products and time points to predict.

Product	Time	Price	Add_spend	Store_traffic	Log_inventory	Competitor_price	Comp._promo	Price_X_Promo	Add_X_Price	Store_size	Cust_demo
1	43931	3.55	490.12	2.78	4.05	3.45	0	0	1650.3	2290	3
1	43932	3.62	510.44	3.01	3.88	3.52	1	3.62	1720.15	2290	3
1	43933	3.48	475.8	2.95	4.1	3.38	0	0	1590.22	2290	3

Download sample datasets from GitHub — also works with CSV files.

Code Walkthrough

Step 1: Initialize

Python

import pandas as pd
from datfid import DATFIDClient

client = DATFIDClient(token="your_DATFID_token")

Step 2: Fit the Model

Python

url_fit = "https://raw.githubusercontent.com/datfid-valeriidashuk/sample-datasets/main/Food_Beverages.xlsx"
df = pd.read_excel(url_fit)

result = client.fit_model(
    df=df,
    id_col="Product",
    time_col="Time",
    y="Revenue",
    current_features="all",
    filter_by_significance=True
)

Step 3: Forecast

Python

url_forecast = "https://raw.githubusercontent.com/datfid-valeriidashuk/sample-datasets/main/Food_Beverages_forecast.xlsx"
df_forecast = pd.read_excel(url_forecast)

forecast = client.forecast_model(df_forecast=df_forecast)

Analysis Results (Model Fit)

Formula

Revenue ~ α1*Intercept + α2*Store_size + α3*Cust_demo + β1*Price + β2*Add_spend + β3*Store_traffic + β4*Log_inventory + β5*Competitor_price + β6*Competitor_promotion

Alpha Estimates (Time-Invariant)

Characteristics constant across time — inherent properties of each product.

Variable	Estimate	T-stat	Interpretation
Intercept	-14,962.3	184.5	Baseline revenue offset (large negative because the model centres on store and demographic effects).
Store_size	+49.9	1,355.8	Each unit of store size adds ~49.9 to revenue.
Cust_demo	+546.3	56.4	Each step up in customer-demographics index adds ~546.3 to revenue.

Beta Estimates (Time-Varying)

Variable	Estimate	T-stat	Interpretation
Price	-332.6	11.6	Each $1 increase in price subtracts ~332.6 in revenue — the headline elasticity.
Add_spend	+53.2	408.3	Each $1 of advertising spend lifts revenue by ~53.2 — strong ROAS.
Store_traffic	-52.9	1.4	Foot traffic moves revenue slightly down in this run, but the effect is not statistically significant.
Log_inventory	+219.6	5.0	A 1-unit increase in log inventory adds ~219.6 to revenue.
Competitor_price	+5,307.7	182.7	Each $1 a competitor raises their price adds ~5,307.7 to our revenue — strong cross-elasticity.
Competitor_promotion	-104.8	7.0	A 1-unit increase in competitor promotion subtracts ~104.8 from revenue.

Model Performance

0.998

R² Overall

1.000

R² Between

0.976

R² Within

432.3

MAE

296,001.0

MSE

Try it yourself: Run this exact analysis in the Free Playground — select "Food Beverages" from the sample datasets and click Run Analysis.