Energy Electricity Forecasting

Forecast energy demand across regions and building types using panel data with 50 entities tracked over 100 time periods.

Dataset Exploration

The Energy Electricity dataset tracks power consumption metrics across different buildings and time periods. Features include price (log-transformed), energy efficiency programs, household activity levels, and building characteristics.

Panel data structure: 50 entities (buildings/regions) tracked over 100 time periods (5,000 total rows, 11 columns).

Fit Data (first 5 rows)

Time	Individual	log(Price)	Energy Eff.	Household Act.	log(P)×Eff.	Building Size	Building Type	Energy Demand
1	1	-0.84	2	3	-1.68	8.23	1	634.54
2	1	-1.1	2	2	-2.2	8.23	1	454.57
3	1	-0.84	2	2	-1.68	8.23	1	470.83
4	1	-0.81	2	2	-1.62	8.23	1	409.56
5	1	-1	2	2	-2	8.23	1	428.04

Forecast Data (first 3 rows)

The forecast file covers future time periods without the Energy Demand target column.

Time	Individual	log(Price)	Energy Eff.	Household Act.	log(P)×Eff.	Private Cust.	Building Size	Building Type
101	1	-1.34	1	2	-1.34	1	8.09	3
102	1	-0.67	1	1	-0.67	1	8.09	3
103	1	-0.66	1	2	-0.66	1	8.09	3

Download sample datasets from GitHub

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/Energy_Electricity.xlsx"
df = pd.read_excel(url_fit)

result = client.fit_model(
    df=df,
    id_col="Individual",
    time_col="Time",
    y="Energy Demand",
    current_features="all",
    filter_by_significance=True
)

Step 3: Forecast

Python

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

forecast = client.forecast_model(df_forecast=df_forecast)

Analysis Results (Model Fit)

Formula

Energy Demand ~ α1*Intercept + α2*Building Size + α3*Private Customer + α4*Building Type + β1*log(Price) + β2*Energy Efficiency Programs + β3*Household Activity + β4*Weekend

Alpha Estimates (Time-Invariant)

Variable	Estimate	T-stat	Interpretation
Intercept	+35.0	0.831	Baseline demand level — not statistically distinguishable from zero on its own (p ≈ 0.41), since building and customer traits absorb most of the level.
Building Size	+32.2	6.0	Each additional unit of building size adds ~32.2 to demand.
Private Customer	+40.4	13.3	Private customers consume ~40.4 more than the baseline business segment.
Building Type	+99.8	63.9	Each step up in building-type index adds ~99.8 to demand — by far the strongest entity-level driver.

Beta Estimates (Time-Varying)

Variable	Estimate	T-stat	Interpretation
log(Price)	-180.3	33.3	A 1-unit increase in log price cuts demand by ~180.3 — the headline price-elasticity effect.
Energy Efficiency Programs	-101.0	24.2	Each program-step trims demand by ~101.0 — efficiency works as advertised.
Household Activity	+50.4	19.7	Each unit of household-activity raises demand by ~50.4.
Weekend	+22.8	7.2	Weekend days run ~22.8 above weekdays in average demand.

Model Performance

0.557

R² Overall

0.897

R² Between

0.305

R² Within

83.8

MAE

10,995.3

MSE

Try it yourself: Run this exact analysis in the Free Playground — select "Energy Electricity" from the sample datasets.