NoGoodGamez

machine learning - based game recommender

1. Retrieving games:

a. at first select games - i prefer to take top gross from vgchartz and form csv with em /data/psn_gamez.csv

b. as we take gamedata from gamespot there might be some url-related issues not covered by function make_gamespot_urlname at /app/fetcher/helpers.php, so add those exceptions to /data/gmz.csv

c. run the retriever - /app/gamez_retrieve.php

d. for updating games based on new gamespot reviews use the updater - /app/gamez_update.php

2. Before working with PredictionIO do not forget to modify readTraining() in DataSource.scala at you PredictionIO server for accepting likes

    val viewEventsRDD: RDD[ViewEvent] = eventsDb.find(
      appId = dsp.appId,
      entityType = Some("user"),
      eventNames = Some(List("like")),//MODIFIED HERE
      // targetEntityType is optional field of an event.
      targetEntityType = Some(Some("item")))(sc)
      // eventsDb.find() returns RDD[Event]
      .map { event =>
        val viewEvent = try {
          event.event match {
            case "like" => ViewEvent(// AND HERE
              user = event.entityId,
              item = event.targetEntityId.get,
              t = event.eventTime.getMillis)
            case _ => throw new Exception(s"Unexpected event ${event} is read.")
          }
        } catch {
          case e: Exception => {
            logger.error(s"Cannot convert ${event} to ViewEvent." +
              s" Exception: ${e}.")
            throw e
          }
        }
        viewEvent
      }.cache()

3.Set data at pio server with PredictionItemsInitial.php

4.I used modified jtinder for dynamic image load you can use the original version jTinder

5.You should update your prediction server addresses

define("KEY", "your_key");
define("ML_SERVER_ADR", "your_server");

Thank you!