Collective Intelligence

The problem being solved is reliable forecasting of various events based on the opinions of expert groups and on statistics of hits and forecast accuracy. The idea is to build a digital tool with a library of connectors to various data sources. The user selects the sources they care about, chooses slices, periods, volume, and the attribute composition of the data. The system pulls the data in via API into a Parquet or CSV table and visualizes it. The datasets are then processed with the bitwise fuzzification method based on user-defined settings and scenarios, after which an algorithm for matching fuzzy data in bit form is launched. Because of the compactness and discrete nature of the representation, the data can be analyzed quickly on the fly. Precision can be configured separately for each indicator.

Experts can work with the time series, compare them using familiar tools, and then produce a forecast and describe why they made it. They can mark the points on the time series that influenced their decision. In the end, based on the forecasts that actually come true, we can identify the winners and compare their reasoning and explanations.