|
|
@@ -626,6 +626,95 @@ PUT _ml/data_frame/analytics/student_performance_mathematics_0.3
|
|
|
<1> The percentage of the data set that is used for training the model.
|
|
|
<2> The seed that is used to randomly pick which data is used for training.
|
|
|
|
|
|
+The following example uses custom feature processors to transform the
|
|
|
+categorical values for `DestWeather` into numerical values using one-hot,
|
|
|
+target-mean, and frequency encoding techniques:
|
|
|
+
|
|
|
+[source,console]
|
|
|
+--------------------------------------------------
|
|
|
+PUT _ml/data_frame/analytics/flight_prices
|
|
|
+{
|
|
|
+ "source": {
|
|
|
+ "index": [
|
|
|
+ "kibana_sample_data_flights"
|
|
|
+ ]
|
|
|
+ },
|
|
|
+ "dest": {
|
|
|
+ "index": "kibana_sample_flight_prices"
|
|
|
+ },
|
|
|
+ "analysis": {
|
|
|
+ "regression": {
|
|
|
+ "dependent_variable": "AvgTicketPrice",
|
|
|
+ "num_top_feature_importance_values": 2,
|
|
|
+ "feature_processors": [
|
|
|
+ {
|
|
|
+ "frequency_encoding": {
|
|
|
+ "field": "DestWeather",
|
|
|
+ "feature_name": "DestWeather_frequency",
|
|
|
+ "frequency_map": {
|
|
|
+ "Rain": 0.14604811155570188,
|
|
|
+ "Heavy Fog": 0.14604811155570188,
|
|
|
+ "Thunder & Lightning": 0.14604811155570188,
|
|
|
+ "Cloudy": 0.14604811155570188,
|
|
|
+ "Damaging Wind": 0.14604811155570188,
|
|
|
+ "Hail": 0.14604811155570188,
|
|
|
+ "Sunny": 0.14604811155570188,
|
|
|
+ "Clear": 0.14604811155570188
|
|
|
+ }
|
|
|
+ }
|
|
|
+ },
|
|
|
+ {
|
|
|
+ "target_mean_encoding": {
|
|
|
+ "field": "DestWeather",
|
|
|
+ "feature_name": "DestWeather_targetmean",
|
|
|
+ "target_map": {
|
|
|
+ "Rain": 626.5588814585794,
|
|
|
+ "Heavy Fog": 626.5588814585794,
|
|
|
+ "Thunder & Lightning": 626.5588814585794,
|
|
|
+ "Hail": 626.5588814585794,
|
|
|
+ "Damaging Wind": 626.5588814585794,
|
|
|
+ "Cloudy": 626.5588814585794,
|
|
|
+ "Clear": 626.5588814585794,
|
|
|
+ "Sunny": 626.5588814585794
|
|
|
+ },
|
|
|
+ "default_value": 624.0249512020454
|
|
|
+ }
|
|
|
+ },
|
|
|
+ {
|
|
|
+ "one_hot_encoding": {
|
|
|
+ "field": "DestWeather",
|
|
|
+ "hot_map": {
|
|
|
+ "Rain": "DestWeather_Rain",
|
|
|
+ "Heavy Fog": "DestWeather_Heavy Fog",
|
|
|
+ "Thunder & Lightning": "DestWeather_Thunder & Lightning",
|
|
|
+ "Cloudy": "DestWeather_Cloudy",
|
|
|
+ "Damaging Wind": "DestWeather_Damaging Wind",
|
|
|
+ "Hail": "DestWeather_Hail",
|
|
|
+ "Clear": "DestWeather_Clear",
|
|
|
+ "Sunny": "DestWeather_Sunny"
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+ ]
|
|
|
+ }
|
|
|
+ },
|
|
|
+ "analyzed_fields": {
|
|
|
+ "includes": [
|
|
|
+ "AvgTicketPrice",
|
|
|
+ "Cancelled",
|
|
|
+ "DestWeather",
|
|
|
+ "FlightDelayMin",
|
|
|
+ "DistanceMiles"
|
|
|
+ ]
|
|
|
+ },
|
|
|
+ "model_memory_limit": "30mb"
|
|
|
+}
|
|
|
+--------------------------------------------------
|
|
|
+// TEST[skip:TBD]
|
|
|
+
|
|
|
+NOTE: These custom feature processors are optional; automatic
|
|
|
+{ml-docs}/ml-feature-encoding.html[feature encoding] still occurs for all
|
|
|
+categorical features.
|
|
|
|
|
|
[[ml-put-dfanalytics-example-c]]
|
|
|
=== {classification-cap} example
|
Lisa Cawley