#!/usr/bin/env python
# coding: utf-8

# # Problem 1
# Apply your skills to classify protein foldType with Decision Tree Classifier
# 
# ## Imports

# In[1]:


from mmtfPyspark.ml import SparkMultiClassClassifier, datasetBalancer                                 
from pyspark.sql import SparkSession
from pyspark.sql.functions import *
import mltoolkit


# ## Configure Spark Session

# In[2]:


spark = SparkSession.builder.appName("Problem-1").getOrCreate()


# ## TODO-1: Read in data from parquet file

# In[3]:


parquetFile = './input_features/'
data = # Your Code Here #


# ## TODO-2: Select alpha, beta, alpha+beta foldtypes

# In[ ]:


data = # Your Code Here #
print(f"Total number of data: {data.count()}")


# ## TODO-3: Downsample data

# In[ ]:


label = 'foldType'

data = # Your Code Here #
print(f"Dataset size (balanced)  : {data.count()}")
    
data.groupby(label).count().show()


# ## TODO-4: Decision Tree Classifier with PySpark

# In[ ]:


from pyspark.ml.classification import DecisionTreeClassifier

dtc = # Your Code Here: Make Decision Tree Classifier Class #
mcc = # Your Code Here: Use MulticlassClassifier wrapper on dtc#
matrics = # Your Code Here: fit data#
for k,v in matrics.items(): print(f"{k}\t{v}")


# ## BONUS: Decision Tree Classifier with sklearn

# In[ ]:


from sklearn.tree import DecisionTreeClassifier

df = # Your Code Here: convert data to Pandas Dataframe #
dtc = # Your Code Here: Make Decision Tree Classifier Class #
mcc = # Your Code Here: Use MulticlassClassifier wrapper on dtc#
matrics = # Your Code Here: fit data#
for k,v in matrics.items(): print(f"{k}\t{v}")


# In[ ]:


spark.stop()