import pandas as pd
import matplotlib.pyplot as plt

# store the data in a pandas series (column in a dataframe)
data = pd.Series([1,2,3,4,5,6])

# compute and print the mean, median, and variance of the data
print("The mean is",data.mean())
print("The median is",data.median())
print("The variance is",data.var())

# find and print the largest number in the data
maxNum = data.max()
print("The biggest value is",maxNum)

# find and print the smallest number in the data
minNum = data.min()
print("The smallest value is",minNum)

# plot the numbers in a histogram that has one bin for each possible number
# bins = range(2,8) tells the program to create bins with boundaries 2,3,4,5,6,7
# These boundaries start at the first number in the range() command, and go up by
# 1 until they are 1 less than the last number.
data.plot.hist(bins = range(minNum,maxNum+2), title="Histogram of data")
plt.show()

# Uncomment for Part B by removing the ''' (make a commment over mutiple lines)
'''
# print basic statistics about the data (number of numbers in the dataset,
# mean, standard deviation (square root of variance), min, 25% percentile, 
# 50% percentile or median, 75% percentile, max)
print(data.describe())

# make a boxplot of the data
data.plot.box(title="Boxplot of data")
# to include the mean: data.plot.box(showmeans=True,title="Boxplot of data")
plt.show()
'''