fix version of tensorflow to 1.* for tensorflow 2 implement a TfKeras…

… Model later

README.md

@@ -221,8 +221,7 @@ df.predict(pmu.Model.load("/tmp/burrito.model")).tail()
 * add Proximity https://stats.stackexchange.com/questions/270201/pooling-levels-of-categorical-variables-for-regression-trees/275867#275867
 
 ## Wanna help?
-* currently I only need binary classification
-    * maybe you want to add a feature for multiple classes
+* for tensorflow 2.x implement a new TfKeras Model
 * for non classification problems you might want to augment the `Summary` 
 * write some tests
 * add different more charts for a better understanding/interpretation of the models

docs/api.rst

@@ -38,6 +38,22 @@ Model
    .. automethod:: __init__
 
 
+SkitModel
+.........
+Simply provide the sklearn model i.e. LogisticRegression along with the features and labels
+
+
+KerasModel
+..........
+.. autoclass:: pandas_ml_utils.Model
+   :members:
+
+   .. automethod:: __init__
+
+NOTE in case of tensorflow backend currently only tensorflow 1.* is supported.
+For tensorflow 2.x we need implement a TfKeras Model.
+
+
 FeaturesAndLabels
 -----------------
 .. autoclass:: pandas_ml_utils.FeaturesAndLabels

pandas_ml_utils/model/models.py

@@ -223,7 +223,9 @@ def __init__(self,
                  **kwargs):
         """
         Keras compatible implementation of :class:`.Model`.
-        :param keras_compiled_model_provider: a callable which provides a compiled ready to fit keras model
+        :param keras_compiled_model_provider: a callable which provides an eventually compiled ready to fit keras model.
+               if the model is not compiled you ned to pass "optimizer" argument as kwargs.
+               NOTE: the tensorflow backend is currently limited to 1.*
         :param features_and_labels: see :class:`.Model`
         :param summary_provider: :class:`.Model`.
         :param epochs: number of epochs passed to the keras fit function

requirements.txt

@@ -2,7 +2,7 @@
 sortedcontainers
 scikit-learn
 statsmodels
-tensorflow
+tensorflow==1.*
 matplotlib
 cachetools
 yfinance