Module 1, Practical 10¶

Part 1¶

Several years ago, researchers compiled a dataset known as the “Copenhagen Network study”. This dataset includes a range of information collected from 700 university students. In this exam, we will focus on the following files:

A list of phone calls: calls.csv
Information about Facebook friendships: fb_friends.csv
Student gender data: genders.txt

The data in these files are structured as follows:

calls.csv

timestamp,caller,callee,duration
184,300,301,121
3920,512,299,670

fb_friends.csv

# user_a,user_b
0,512
0,263
0,525

gender.txt

0_M 2_M 3_M 4_M 5_M ...

Write a program that performs the following tasks:

Find the Longest Call: Identify the call with the longest duration from the dataset.
Check Facebook Friendship: Determine if the caller and receiver in the longest call are friends on Facebook.
Display Genders: Print the gender of both students involved in the longest call.
Discretize Interaction Durations: Convert the duration of each interaction from seconds into hourly units. Any interaction that occurs within one hour should be rounded up to 1 (for example, 45 minutes or 3600 seconds both discretize to 1 hour).

Input: The program should take in the file CALLS.

    time  id_a id_b duration
[
    [  184,   300,   301,   121],
    [ 3920,   512,   299,   670],
    [ 5623,   301,   300,   504],
    [ 9252,   401,   457,    -1],
    [15466,   512,     0,     5],
    [15497,   512,     0,    28],
    [26400,    19,    47,   619],
    [31312,   687,   310,    11],
    [36265,   300,   301,    74],
    [37049,   634,   681,    20],
   ...,
]

output CALLS

    time  id_a id_b duration
[
    [  0,   300,   301,   121],
    [  1,   512,   299,   670],
    [  1,   301,   300,   504],
    [  2,   401,   457,    -1],
    [  4,   512,     0,     5],
    [  4,   512,     0,    28],
    [  7,    19,    47,   619],
    [  8,   687,   310,    11],
    [ 10,   300,   301,    74],
    [ 10,   634,   681,    20],
   ...,
]

then plot those interactions, you should obtain a plot like the one bellow! pay attention to x and y labels

NOTE you should define a function for each point.

Show/Hide Solution

[ ]:

import pandas as pd
import matplotlib.pyplot as plt
import numpy as np

def get_longest_call(calls):
    longest_call = calls.loc[calls['duration'].idxmax()]
    id_caller, id_callee = longest_call.caller,longest_call.callee
    return id_caller,id_callee


def get_friends_fb(fb):
    fb_friends = []
    for _,i in fb.iterrows():
        user_a = i[0]
        user_b = i[1]
        fb_friends.append((user_a,user_b))

    return fb_friends

def get_gender(gender,id_caller,id_callee):
    print("gender of ",id_callee," = ",gender[id_callee])
    print("gender of ",id_caller," = ",gender[id_caller])

def get_sms_count_dict(sms):
    sms_count = dict()
    for _,i in sms.iterrows():
        _,sender,reciver = i

        if sender > reciver:
            tmp = sender
            sender = reciver
            reciver = tmp

        if (sender,reciver) in sms_count:
            sms_count[(sender,reciver)] += 1
        else:
            sms_count[(sender,reciver)] = 1

    return sms_count


def get_tot_messages(sms_count,fb_friends):
    tot_messages = 0
    for k,v in sms_count.items():
        id_a, id_b = k

        if not (id_a,id_b) in fb_friends:
            if not (id_b,id_a) in fb_friends:
                tot_messages += v
    return tot_messages

def load_gender(path):
    f = open(path, "r")
    data = f.read()
    f.close()
    res = dict()
    for el in data.split(" "):
        tmp = el.split("_")
        idx = tmp[0]
        gender = tmp[1]
        res[int(idx)] = gender
    return res

def get_discretized_times(calls_small):
    times_call = []


    for t,a,b,d in calls_small.to_numpy():
        times_call.append(int(t/(60*60)))

    return times_call

def my_plot(times_call):

    t_c,count_c = np.unique(times_call,return_counts=True)

    x_ticks = []
    x_labels = []
    for i in range(int((max(t_c)/24)+1)):
        x_ticks.append(i*24)
        x_labels.append("Day "+str(i+1))

    plt.figure(figsize=(15,4))
    plt.plot(t_c,count_c,label="calls",color="blue")
    plt.ylabel("Number of interactions")
    plt.xticks(x_ticks, x_labels,rotation=45)
    plt.title("Calls")
    plt.legend()
    plt.savefig("nb_interactions.png")
    plt.show()



# load data
fb = pd.read_csv("DTU/fb_friends.csv")
calls = pd.read_csv("DTU/calls.csv")


# 1)
id_caller, id_callee = get_longest_call(calls)
print("1)","the longest call is:",id_caller, "and", id_callee)

# 2)
fb_friends = get_friends_fb(fb)
if (id_callee,id_caller) in fb_friends or (id_caller,id_callee) in fb_friends:
    print("2)",id_callee," and ",id_caller," are friend of FB")

# 3)
print("3)")
gender = load_gender("DTU/gender.txt")
get_gender(gender,id_caller,id_callee)

# 4)
times_call = get_discretized_times(calls)
my_plot(times_call)

Running cells with 'Python 3.10.6' requires the ipykernel package.

Run the following command to install 'ipykernel' into the Python environment.

Command: '/usr/bin/python3 -m pip install ipykernel -U --user --force-reinstall'

[5]:

import pandas as pd
import matplotlib.pyplot as plt
import numpy as np




def max_call(call):
    m = call.duration.max()
    cd = call[call.duration == m]
    return cd.caller.item(), cd.callee.item()

def fun2(fb,a,b):

    for idx,row in fb.iterrows():
        print(row)
        ca = row["# user_a"]
        cb = row["user_b"]

        if a == ca and b == cb:
            return True
        elif a == cb and b == ca:
            return True

    return False

def fun2_1(fb,a,b):
    d = dict()

    for idx,row in fb.iterrows():
        ca = row["# user_a"]
        cb = row["user_b"]

        d[(ca,cb)] = 1

    if (a,b) in d:
        return True
    elif (b,a) in d:
        return True
    else:
        return False

def fun3(a,b):

    with open("DTU/gender.txt","r") as f:
        gender_in = f.readlines()[0]

    gender = dict()

    for i in gender_in.split(" "):
        idx = i.split("_")[0]
        gen = i.split("_")[1].replace("\n","")


        gender[int(idx)] = gen
    print(gender[a],gender[b])

def fun4(call):
    new = []
    for a,b,c,d in call.to_numpy():
        new_a = int(a/3600)
        new.append([new_a,b,c,d])
    res = pd.DataFrame(new,columns=call.columns)

    return res

def fun4(call):
    call.timestamp = call.timestamp//3600
    return call



call = pd.read_csv("DTU/calls.csv")
fb = pd.read_csv("DTU/fb_friends.csv")


a,b = max_call(call)

print(fun2_1(fb,b,a))
print(a,b)
fun3(a,b)
call2 = fun4(call)


v,h = np.unique(call2.timestamp,return_counts=True)


plt.figure(figsize=(10,4))
plt.plot(v,h,label="calls")
plt.ylabel("nb of interactions")
plt.title("calls")

x_ticks = []
x_labels = []
for i in range(int((max(v)/24)+1)):
    x_ticks.append(i*24)
    x_labels.append("Day "+str(i+1))

plt.xticks(x_ticks, x_labels,rotation=45)

plt.legend()
plt.show()

True
136 137
M F

[ ]: