Post-Score Recommender Offers Plugin
The Post Score Recommender plugin is used to customize the output of a predictor by formatting the output into a JSON object. This plugin is designed to be used in conjunction with other post-predict plugins to customize the output of the predictor. The Post Score Recommender plugin is a simple plugin that is designed to be used in conjunction with other post-predict plugins to customize the output of the predictor.
How does it work?
The Post Score Recommender Offer plugin loops over the eligible offers as per the offer matrix associated with the project. Note that this Plugin requires an Offer Matrix to be configured with a specific data format.
Java Code
This Java code follows the default standard as outlined in the Basic post score plugin. Note that this class obtains the Offer Matrix from the pre-defined loaded data as defined here: offerMatrix = params.getJSONArray("offerMatrix");.
This recommender system ranks and filters product offers based on the model’s predicted probabilities. It is designed for use with a single predictive model trained on offer_id as the response column. The input offerMatrix is expected to contain all available offers, each with pricing details.
The key method, getPostPredict, processes the model’s prediction output along with input parameters to:
- Retrieve probabilities per offer from model output.
- Match these probabilities against offer IDs or names from the input offer matrix.
- Compute a modified_offer_score using the formula: probability * (offer_value - offer_cost)
- Sort the offers based on this score.
- Return the top-ranked offers for further downstream usage.
The following is an example of the Post Score Recommender Offer plugin implemented in Java:
package com.ecosystem.plugin.customer;
import com.datastax.oss.driver.api.core.CqlSession;
import com.ecosystem.utils.DataTypeConversions;
import com.ecosystem.utils.JSONArraySort;
import hex.genmodel.easy.EasyPredictModelWrapper;
import com.ecosystem.utils.log.LogManager;
import com.ecosystem.utils.log.Logger;
import org.json.JSONArray;
import org.json.JSONObject;
/**
* recommender_smp - Single model for all products with Offermatrix
* Multiclass classifier trained on offer_name response column, offer matrix need to have all the offers loaded with offer_price.
*/
public class PostScoreRecommenderOffers extends PostScoreSuper {
private static final Logger LOGGER = LogManager.getLogger(PostScoreRecommenderOffers.class.getName());
public PostScoreRecommenderOffers() {
}
/**
* Pre-post predict logic
*/
public void getPostPredict () {
}
/**
* getPostPredict
*
* @param predictModelMojoResult Result from scoring
* @param params Params carried from input
* @param session Session variable for Cassandra
* @return JSONObject result to further post-scoring logic
*/
public static JSONObject getPostPredict(JSONObject predictModelMojoResult, JSONObject params, CqlSession session, EasyPredictModelWrapper[] models) {
double startTimePost = System.nanoTime();
try {
/** Value obtained via API params */
JSONObject work = params.getJSONObject("in_params");
double in_balance = 1000.0;
if (work.has("in_balance"))
in_balance = DataTypeConversions.getDouble(work, "in_balance");
else
LOGGER.info("getPostPredict:I001aa: No in_balance specified, default used. (1000.00)");
JSONArray sortJsonArray = new JSONArray();
JSONArray finalOffers = new JSONArray();
/* Setup JSON objects for specific prediction case */
JSONObject featuresObj = predictModelMojoResult.getJSONObject("featuresObj");
if (predictModelMojoResult.has("ErrorMessage")) {
LOGGER.error("getPostPredict:E001a:" + predictModelMojoResult.get("ErrorMessage"));
return null;
}
JSONArray offerMatrix = new JSONArray();
if (params.has("offerMatrix"))
offerMatrix = params.getJSONArray("offerMatrix");
JSONObject domainsProbabilityObj = predictModelMojoResult.getJSONObject("domainsProbabilityObj");
try {
String label = predictModelMojoResult.getJSONArray("label").getString(0).trim();
JSONArray domains = predictModelMojoResult.getJSONArray("domains");
} catch (Exception e) {
LOGGER.error("getPostPredict:E001b:Model could not be loaded, check deployment path: " + e);
}
int offerIndex = 0;
int explore = (int) params.get("explore");
/** Select top items based on number of offers to present */
for (int i = 0; i < offerMatrix.length(); i++) {
JSONObject singleOffer = offerMatrix.getJSONObject(i);
JSONObject finalOffersObject = new JSONObject();
double offer_value = 1.0;
if (singleOffer.has("offer_price"))
offer_value = DataTypeConversions.getDouble(singleOffer, "offer_price");
if (singleOffer.has("price"))
offer_value = DataTypeConversions.getDouble(singleOffer, "price");
double offer_cost = 1.0;
if (singleOffer.has("offer_cost"))
offer_cost = singleOffer.getDouble("offer_cost");
if (singleOffer.has("cost"))
offer_cost = singleOffer.getDouble("cost");
String offer_id = "";
String offer_name = "";
if (domainsProbabilityObj.has(singleOffer.getString("offer_id").trim())) {
offer_id = singleOffer.getString("offer_id").trim();
} else if (domainsProbabilityObj.has(singleOffer.getString("offer").trim())) {
offer_id = singleOffer.getString("offer").trim();
} else if (domainsProbabilityObj.has(singleOffer.getString("offer_name").trim())) {
offer_name = singleOffer.getString("offer_name").trim();
} else {
LOGGER.error("offerRecommender:E002-1: " + params.get("uuid") + " - Not available (offer_id, offer, offer_name from probabilities): " + singleOffer.getString("offer_name"));
}
double p = 0.0;
if (domainsProbabilityObj.has(offer_id.trim())) {
offer_id = singleOffer.getString("offer_id").trim();
p = domainsProbabilityObj.getDouble(offer_id);
} else {
LOGGER.error("offerRecommender:E002-1: " + params.get("uuid") + " - Not available: " + singleOffer.getString("offer_name"));
}
double modified_offer_score = 1.0;
modified_offer_score = p * ((double) offer_value - offer_cost);
finalOffersObject.put("offer", offer_id);
finalOffersObject.put("offer_name", offer_name);
finalOffersObject.put("offer_name_desc", offer_name + " - " + i);
/** process final */
// double p = domainsProbabilityObj.getDouble(label);
finalOffersObject.put("score", p);
finalOffersObject.put("final_score", p);
finalOffersObject.put("modified_offer_score", modified_offer_score);
finalOffersObject.put("offer_value", offer_value); // use value from offer matrix
finalOffersObject.put("price", offer_value);
finalOffersObject.put("cost", offer_cost);
finalOffersObject.put("uuid", params.get("uuid"));
finalOffersObject.put("p", p);
finalOffersObject.put("explore", explore);
/** Prepare array before final sort */
finalOffers.put(offerIndex, finalOffersObject);
offerIndex = offerIndex + 1;
}
sortJsonArray = JSONArraySort.sortArray(finalOffers, "modified_offer_score", "double", "d");
predictModelMojoResult.put("final_result", sortJsonArray);
/** Select the correct number of offers */
predictModelMojoResult = getTopScores(params, predictModelMojoResult);
} catch (Exception e) {
e.printStackTrace();
LOGGER.error(e);
}
/** Top scores from final_result */
predictModelMojoResult = getTopScores(params, predictModelMojoResult);
double endTimePost = System.nanoTime();
LOGGER.info("PostScoreRecommenderOffers:I001: time in ms: ".concat( String.valueOf((endTimePost - startTimePost) / 1000000) ));
return predictModelMojoResult;
}
}