Overall zero search results rate

ARIMA Forecast

BSTS Forecast

Prophet Forecast

Wikidata Query Service SPARQL endpoint usage

ARIMA Forecast

BSTS Forecast

Prophet Forecast


Link to this dashboard: https://discovery.wmflabs.org/forecasts/ | Page is available under CC-BY-SA 3.0 | Code is licensed under MIT | Part of Discovery Dashboards | Forecasting Code available as part of this repository

Overall zero results rate

Proportion of searches that yield zero results.

We studied multiple autoregressive integrated moving average (ARIMA) model models using the explorer app, and settled on ARIMA(2,1,2)x(1,0,0) w/ period of 7 as the best one – decided by Akaike Information Criterion (AIC).

We also have two competing models:

Note that the “Reportupdater” annotation refers to when we switched our data retrieval and processing codebase to Wikimedia Analytics' Reportupdater infrastructure. See T150915 for more details.


Link to this dashboard: https://discovery.wmflabs.org/forecasts/ | Page is available under CC-BY-SA 3.0 | Code is licensed under MIT | Part of Discovery Dashboards | Forecasting Code available as part of this repository

Wikidata Query Service homepage traffic

The WDQS homepage (query.wikidata.org) is the best way to use WDQS. It includes examples and a host of other user-friendly features.

We studied multiple autoregressive integrated moving average (ARIMA) model models using the explorer app, and settled on ARIMA(1,1,2)x(1,1,0) w/ period of 7 as the best one – decided by Akaike Information Criterion (AIC).

We also have two competing models:

Note that the “Reportupdater” annotation refers to when we switched our data retrieval and processing codebase to Wikimedia Analytics' Reportupdater infrastructure. See T150915 for more details.


Link to this dashboard: https://discovery.wmflabs.org/forecasts/ | Page is available under CC-BY-SA 3.0 | Code is licensed under MIT | Part of Discovery Dashboards | Forecasting Code available as part of this repository

Wikidata Query Service SPARQL endpoint usage

WDQS API can be used by third parties to query Wikidata using SPARQL.

We studied multiple autoregressive integrated moving average (ARIMA) model models using the explorer app, and settled on ARIMA(1,1,1)x(1,0,0) w/ period of 7 as the best one – decided by Akaike Information Criterion (AIC).

We also have two competing models:

Note that the “Reportupdater” annotation refers to when we switched our data retrieval and processing codebase to Wikimedia Analytics' Reportupdater infrastructure. See T150915 for more details.


Link to this dashboard: https://discovery.wmflabs.org/forecasts/ | Page is available under CC-BY-SA 3.0 | Code is licensed under MIT | Part of Discovery Dashboards | Forecasting Code available as part of this repository