Two forecasts. Zero historical data. Here’s how I built them.

1. 1Identify the problem domain and understand the requirements
2. 2Determine the type of forecast needed (e.g. time series, regression)
3. 3Explore alternative data sources (e.g. similar products, markets, or industries)
4. 4Apply machine learning techniques (e.g. transfer learning, meta-learning) to build the forecast model
5. 5Evaluate and refine the model using metrics (e.g. mean absolute error, mean squared error)

Data scientists, analysts, and product managers can benefit from this skill to make informed decisions in the absence of historical data

💡 Forecasts can be built without historical data by leveraging alternative data sources and machine learning techniques

📊 No historical data? No problem! Learn how to build forecasts from scratch #datascience #forecasting