Implement prediction model - Githubissues

anebz / boulder

Occupancy tracker for bouldering gyms

http://boulder.anebz.eu

7 stars 3 forks source link

Implement prediction model #13

Closed anebz closed 2 years ago

anebz commented 3 years ago

backend

[x] Write code to train model with all existing data
[x] ~Listen on some port for the frontend's signal~
[x] Predict and return the predicted values
[x] ~FastAPI?~
[ ] Lambda function that runs ev day, ev few days. get new data, add fit again, write model to s3 again

model.save_model('0001.model')
---
model2 = xgb.Booster()
model2.load_model('0001.model')
res = model.predict(X)

frontend

[x] ~Connect with backend container~
[x] Load model from S3
[ ] Show predicted values in plot
[ ] Every 15mins, update plot with actual value and make prediction again. Maybe make predictions 1-2h into the future

visualization small fix

[x] Don't plot data after 23:00
[x] Round up queue to the next integer
[x] Delete queue column from dataset, add it into occupancy