ADD:win_predict.bat for windows

bin/win_predict.bat

@@ -0,0 +1,28 @@
+echo on
+@Rem model name(linear, fm, ffm, gbdt, gbmlr, gbsdt, gbhmlr, gbhsdt, multiclass_linear)
+set model_name=???
+
+set config_path=config/model/%model_name%.conf
+
+@Rem data file for predicting
+set file_name=???
+
+@Rem train/test line python transform switch & script
+set transform="false"
+set transform_script_path=bin/transform.py
+
+@Rem result save mode: PREDICT_RESULT_ONLY, LABEL_AND_PREDICT, PREDICT_AS_FEATURE
+set resultSaveMode=PREDICT_RESULT_ONLY
+set resultFileSuffix=_%model_name%_%resultSaveMode%
+
+@Rem max error data format tolerate number
+set max_error_tol=100
+
+set eval_metric="auc,mae"
+
+@Rem value or leafid
+set predict_type=value
+
+@start /b cmd /c java -Xmx1000m -XX:-OmitStackTraceInFastThrow -cp lib/* -Dlog4j.configuration=file:config/log4j.properties com.fenbi.ytklearn.predictor.Predicts %config_path% %model_name% %file_name% %transform% %transform_script_path% %resultSaveMode% %resultFileSuffix% %max_error_tol% %eval_metric% %predict_type%
+
+pause

docs/data_format.md

@@ -144,7 +144,7 @@ In default, ```bin/transform.py``` is not used, and you can set ```transform="tr
 
 ### **Data Format Transform Tool** ###
 
-Ytklearn provides the script[libsvm_convert_2_ytklearn.sh](../bin/libsvm_convert_2_ytklearn.sh) to transform data format from libsvm to ytklearn. It will **generate a new dataset file** of ytklearn format and the weight of each sample is one.
+Ytklearn provides the script[libsvm_convert_2_ytklearn.sh](../bin/libsvm_convert_2_ytklearn.sh) ([win_libsvm_convert_2_ytklearn.sh](../bin/win_libsvm_convert_2_ytklearn.sh) for windows) to transform data format from libsvm to ytklearn. It will **generate a new dataset file** of ytklearn format and the weight of each sample is one.
 
 | Task                                     | Desc                                     | Example                     | Original Format                          | Converted Format                         |
 | ---------------------------------------- | ---------------------------------------- | --------------------------- | ---------------------------------------- | ---------------------------------------- |

docs/evaluation_metrics.md

@@ -66,7 +66,7 @@ If you provide labels,  you can set ``eval_metric`` to perform evaluations. Ytk-
 
  You can see the details in ``log/info.log``, including evaluations and a file path of prediction results.
 
-The following script is [predict.sh](../bin/predict.sh).
+The following script is [predict.sh](../bin/predict.sh) ([win_predict.sh](../bin/win_predict.sh) for windows).
 
 ```shell
 #!/usr/bin/env bash