1_DS_analysis.py

# # Load the data

from pyspark.sql import SparkSession
from pyspark.sql.types import Row, StructField, StructType, StringType, IntegerType

spark = SparkSession.builder\
    .appName("Import Wine Table")\
    .config("spark.yarn.access.hadoopFileSystems","s3a://ml-field/demo/wine/")\
    .getOrCreate()

wine_data_raw = spark.sql("SELECT * FROM `default`.`wine`")
wine_data_raw.show(3) 

# ### Basic DataFrame operations
# Dataframes essentially allow you to express sql-like statements. 
# We can filter, count, and so on. 
# Documentation - (http://spark.apache.org/docs/latest/sql-programming-guide.html#dataframe-operations)

"number of lines in dataset : {:d}".format(wine_data_raw.count())

# ### Spark SQL - manipulate data as if it was a table 
wine_data_raw.createOrReplaceTempView("wine")
spark.sql("select distinct(Quality), count(*) from wine GROUP BY Quality").show()


# ### Remove invalid data
wine_data = wine_data_raw.filter(wine_data_raw.Quality != "1")
total_wines = wine_data.count()
good_wines = wine_data.filter(wine_data.Quality == 'Excellent').count()
good_wines = wine_data.filter(wine_data.Quality == 'Poor').count()

"Wines total: {}, Good : {}, Poor : {}".format(total_wines,good_wines,good_wines )


# # 2. Data visualisation ( using mathplotlib and Seaborn)
# ## Feature Visualization
# 
# The data vizualization workflow for large data sets is usually:
# 
# * Sample data so it fits in memory on a single machine.
# * Examine single variable distributions.
# * Examine joint distributions and correlations.
# * Look for other types of relationships.
# 
# [DataFrame#sample() documentation](http://people.apache.org/~pwendell/spark-releases/spark-1.5.0-rc1-docs/api/python/pyspark.sql.html#pyspark.sql.DataFrame.sample)

# ### Note: toPandas() => brings data localy !!!
sample_data = wine_data.sample(False, 0.5, 83).toPandas()
sample_data.transpose().head(21)

spark.stop()


# ## Feature Distributions
# 
# We want to examine the distribution of our features, so start with them one at a time.
# 
# Seaborn has a standard function called [dist()](http://stanford.edu/~mwaskom/software/seaborn/generated/seaborn.distplot.html#seaborn.distplot) that allows us to easily examine the distribution of a column of a pandas dataframe or a numpy array.

get_ipython().magic(u'matplotlib inline')
import matplotlib.pyplot as plt
import seaborn as sb

sb.distplot(sample_data['Alcohol'], kde=False)

# We can examine feature differences in the distribution of our features when we condition (split) our data.
# 
# [BoxPlot docs](http://stanford.edu/~mwaskom/software/seaborn/generated/seaborn.boxplot.html)


sb.boxplot(x="Quality", y="Alcohol", data=sample_data)

# ## Joint Distributions
# 
# Looking at joint distributions of data can also tell us a lot, particularly about redundant features. [Seaborn's PairPlot](http://stanford.edu/~mwaskom/software/seaborn/generated/seaborn.pairplot.html#seaborn.pairplot) let's us look at joint distributions for many variables at once.


example_numeric_data = sample_data[["fixedAcidity", "volatileAcidity",
                                       "citricAcid", "residualSugar", "Quality"]]
sb.pairplot(example_numeric_data, hue="Quality")