recommend-emr-dataprocess/src/main/java/examples/myUtils/FissionConvert.java

package examples.myUtils;

import com.alibaba.fastjson.JSONObject;

import java.util.*;

public class FissionConvert {
    private static final double plusSmooth = 3.0;
    private static final Set<String> realSet = new HashSet<>(Arrays.asList("hour", "group_n", "user_launchs", "user_shares", "user_return_1_uv", "user_return_n_uv"));
    private static final Set<String> hotSceneSet = new HashSet<>(Arrays.asList("1008", "1007", "1074", "1058", "1067", "1010", "1095"));

    public static JSONObject getFeature(String dt, Map<String, String> record, int accurate) {
        Map<String, Double> featMap = new HashMap<>();
        try {
            String dtKey = String.format("dt@%s", dt);
            featMap.put(dtKey, 1.0);
            Map<String, String> feature = ConvertUtils.getRecordCol(record, "feature");
            for (Map.Entry<String, String> entry : feature.entrySet()) {
                String key = entry.getKey();
                String value = entry.getValue();
                if (realSet.contains(key)) {
                    double featValue = Double.parseDouble(value);
                    featMap.put(key, featValue);
                } else {
                    if ("scene".equals(key)) {
                        String featKey;
                        if (hotSceneSet.contains(value)) {
                            featKey = String.format("%s@%s", key, value);
                        } else {
                            featKey = String.format("%s@%s", key, "other");
                        }
                        featMap.put(featKey, 1.0);
                    } else {
                        String featKey = String.format("%s@%s", key, value);
                        featMap.put(featKey, 1.0);
                    }

                }
            }
            // rate
            double launchs = Double.parseDouble(feature.getOrDefault("user_launchs", "0"));
            double shares = Double.parseDouble(feature.getOrDefault("user_shares", "0"));
            double return_1_uv = Double.parseDouble(feature.getOrDefault("user_return_1_uv", "0"));
            double return_n_uv = Double.parseDouble(feature.getOrDefault("user_return_n_uv", "0"));
            if (launchs > 0) {
                double rov_1 = FeatureUtils.plusSmooth(return_1_uv, launchs, plusSmooth);
                double rov_n = FeatureUtils.plusSmooth(return_n_uv, launchs, plusSmooth);
                double str = FeatureUtils.plusSmooth(shares, launchs, plusSmooth);
                featMap.put("rov_1", rov_1);
                featMap.put("rov_n", rov_n);
                featMap.put("str", str);
            }
            if (shares > 0) {
                double ros_1 = FeatureUtils.plusSmooth(return_1_uv, shares, plusSmooth);
                double ros_n = FeatureUtils.plusSmooth(return_n_uv, shares, plusSmooth);
                featMap.put("ros_1", ros_1);
                featMap.put("ros_n", ros_n);
            }
        } catch (Exception e) {
            e.printStackTrace();
        }
        return ConvertUtils.filterAndTruncate(featMap, accurate);
    }
}