import pandas as pd

df = pd.read_csv(r'./data/Ex2aggFlows_team13.csv')

dataLength = len(df)

mean_packet_count = df['packetTotalCount'].mean()
mean_unique_source = df['distinct(sourceIPAddress)'].mean()
mean_num_bytes = df['ipTotalLength'].mean()

print("Mean `packetTotalCount`: {}".format(round(mean_packet_count, 3)))
print("Mean `distinct(sourceIPAddress)`: {}".format(round(mean_unique_source, 3)))
print("Mean `ipTotalLength`: {}".format(round(mean_num_bytes, 3)))

median_packet_count = df['packetTotalCount'].median()
median_unique_source = df['distinct(sourceIPAddress)'].median()
median_num_bytes = df['ipTotalLength'].median()

print("Median `packetTotalCount`: {}".format(round(median_packet_count, 3)))
print("Median `distinct(sourceIPAddress)`: {}".format(round(median_unique_source, 3)))
print("Median `ipTotalLength`: {}".format(round(median_num_bytes, 3)))

std_packet_count = df['packetTotalCount'].std()
std_unique_source = df['distinct(sourceIPAddress)'].std()
std_num_bytes = df['ipTotalLength'].std()

print("Std `packetTotalCount`: {}".format(round(std_packet_count, 3)))
print("Std `distinct(sourceIPAddress)`: {}".format(round(std_unique_source, 3)))
print("Std `ipTotalLength`: {}".format(round(std_num_bytes, 3)))