RazvanO2/Household_Electricity_Forecasting
Folders and files
| Name | Name | Last commit date | ||
|---|---|---|---|---|
Repository files navigation
The data in forecasting_test.csv consists of data on household electricity consumption. The variables are as follows: - hh-id: unique household id - year: 2010 and 2011 - month: 4-8 - zipcode: anonymized zipcode in which home is located - mozip: location variable derived from the interaction of zipcode with year and month. proxies for local humidity/temperature - lusage: log(kwh) log of monthly electricity consumption - lusage1-6: log(kwh) for April - September of 2009 (ie pre-sample period) - children: household has children - hhsize2-5plus: household size - income2-9: income categories <$20k, $20-30k, $30-40k, $40-50k, $50-75k, $75-100k, $100-125k, >$125k - owner: resident owns home - size: size of the residence (in sqft) Note: household electricity usage is based on billing records household demographics were purchased from a third party data aggregator Objective: The aim of this mini-project is to develop a forecasting model to predict future electricity usage (lusage) for each household based on the available data. 1. Prepare Data: Create time-series features, considering past usage, trends, and seasonal effects. 2. Modeling: Build and evaluate forecasting models. 3. Forecasting: Generate predictions with confidence intervals to assess uncertainty. Note on Missing Data & Imputation: * Handle any missing data, as it may impact model accuracy. * Consider the impact of different imputation methods on the forecasting accuracy.