What are predictions?

Predictions are statements about unfinished activities based on what happened to finished activities. We focus on predictions associated with the completion date of unfinished activities and how that completion date can change, compared to the planned completion date. 

How does prediction work?

Once you upload a schedule, we look at each activity and collect both planned and actualised data. Typically, all activities have planned data associated with them - this is usually information around their planned duration and start/end date. Activities which are completed (and sometimes, when in progress) also have actualised data, which reflects what actually happened.

We then focus on predicting actualised data of unfinished activities based on the actual data of finished activities. This prediction is weighted based on the similarity between the unfinished and finished activities. 

Examples of planned data include target duration, target start/end date, target WBS and target resources. Similarity, actualised data includes actual duration, actual start/end date, actual WBS and actual resources. 

How do you know which activities are similar?

We consider two activities as being similar if their planned data are similar. This similarity is calculated using a similarity score, which weights the level of similarity between the planned data of the two activities. The more common properties two activities have, the more similar they are deemed to be. 

Does your AI improve with more data? 

Yes, the more schedules you upload for a specific project the more refined the predictions become. 

The way this works is that the more data you upload, the more information our AI can see around activities that are now completed (and their completion performance). This helps our AI to learn more about your project, and in turn refine its predictions. 

I only have a baseline schedule for now. Can I still get some predictions? 

Yes, you can still leverage the power of predictive AI. Our technology will automatically see that no actualised data exists within the schedule you have uploaded. Based on that observation, the AI will fall back to the curated Nodes & Links database and use that to provide indicative predictions until you upload your own actualised data.