Axis in Matplotlib

Q&A Non-DQ Courses
, ,
#1

Hi DQ-Community,

i work on a little Project to boost my Data Visualization skills with Python and now i’m stuck.
I hope somebody can give me an answer to my question:

I want to visualize the amount of E-Mails during a specific period. I decided to use Matplotlib for this Project. In the x-axis i add the calender week and i want to add the amount of not yet read E-Mails in the y-axis. I have a column with “read Mails” and “Not-read Mails”.
Now my question is:

Is it possible to count the Amount of the “Not-Read Mails” with Pandas and assign the value to the y-axis? How can i do this in Pandas?

Thanks for your help.

Simo

#2

@Simo: Do you mind uploading your dataset and a screenshot of how your data looks like in excel? Thanks

#3

Assumption : your data is a time series (Pandas series), your index is the date put in correct datetime format, and the values are the number of not read emails . Then:

ar = np.zeros(shape=(7,),dtype=np.float32)
for i in range(len(data)):        
     d=data.index[i].weekday()
     ar[d]+=data.iloc[i]
1 Like
#4

Hi @masterryan.prof @WilfriedF i uploaded the anonymized Data, because i see that i’m even not able to create a basic Time Series :confused: The Structure is the same.

I wanted to add on the y-axis the total count of the Mails with the Lable “New” in the Column State_Now and on the x-axis the Calendar week - but it seems that also this is not working with my solution.

Here is my Code:

import pandas as pd
from pandas import DataFrame, Series
from pathlib import Path
import matplotlib.pyplot as plt
pd.set_option("display.max_rows", None, "display.max_columns", None)

input_file = Path.cwd() / 'data' / 'raw' / 'Pandas_Example.xlsx'
df = pd.read_excel(input_file)
df['Week_of_Year'] = pd.to_datetime(df['Week_of_Year'])

#plt.plot()
#plt.show()
#first_twelve = df[0:12]
plt.plot(df['Week_of_Year'], df['State_Now'] =="New")
plt.show()

As you can see on the Comments i tried to adapt the learning from DQ to my own Dataset, but it doesn’t work.

What i have to do to create a nice Time Series over the Weeks?

Thanks as always guys
Simo

Example_Pandas.xlsx (193.0 KB) ,

#5 
df = df.set_index("Send_Time")
df.index = pd.to_datetime(pd.to_datetime(df.index).date)

does it work ?

#6

Hi @WilfriedF,

i have to use the last colum Week_of_Year. Your solution refers to the individual Mail time, right?
I’m not sure if it was necessary to do pd.to_datetime in my case i did it only because i have seen it in the DQ-Course.

#7

Sorry I first thought you wanted weekday on x-axis, but if you want week num of the year, it’s easiest indeed:

data = df.set_index('Week_of_Year')
ar = np.zeros(shape=(52,),dtype=np.float32)
for i in range(len(data)):        
     d=data.index[i]
     if data["State_Now"].iloc[i] == "New":
         ar[d]+=1
#8

Where the np. at zeros and float32 refers to?
Do i have to Import someting?
It runs on an error here: NameError: name ‘np’ is not defined

#9

Sorry this is numpy indeed

import numpy as np

then you plot ar array

Sure you can do it without creating a numpy array but I usually proceed like this

dtype float32 is not necessary here since we deal with int numbers, but it doesn’t matter

#10

Thanks i imported numpy :slight_smile:But the result doesn’t make sense.

image

:confused:

#11

Wait looking at your excel, I see you will have a problem since there is a 53 week number
So you need to adapt the code

d=data.index[i]-1
if data["State_Now"].iloc[i] == "New":
   ar[d]+=1
#12

When you plot:
x = range(0,52)
y = ar

plt.plot(x,y)

still not making any sense ?

your array should have values > 1

#13

I did that but the chart it still confusing.
Do i have to do manually something to the axis?
I expected a Line which will go from left to the right over the Calendar weeks showing the progress/trend of the amont of the new Mails. I thought that my raw Data is good enough for this requirement. In the Tutorials it looks so easy :slight_smile:

EDIT:

Let me check with the plots.

#14

Captura

check it, there are very few “New” values

only 7

#15

Can you please post the whole code? It seems that I did something wrong with the order of the single statements we discussed. My result looks still like the screenshot, very confusing.

#16

If you want accumulation of “New” values among time, this is another story you will need to use cumsum for example

#17 
data = df.set_index('Week_of_Year')
ar = np.zeros(shape=(52,),dtype=np.float32)
for i in range(len(data)):        
    d=data.index[i]-1
    if data["State_Now"].iloc[i] == "New":
        ar[d]+=1
plt.plot(ar)

Look with cumsum:

data = df.set_index('Week_of_Year')
ar = np.zeros(shape=(52,),dtype=np.float32)
for i in range(len(data)):        
    d=data.index[i]-1
    if data["State_Now"].iloc[i] == "New":
        ar[d]+=1
plt.plot(np.cumsum(ar))
plt.xlabel("week")

Captura

#18

Yeah, thats great. Let me try it with cumsum on my site.
So the Whole Code Should be:

import pandas as pd
import numpy as np
from pandas import DataFrame, Series
from pathlib import Path
import matplotlib.pyplot as plt
pd.set_option("display.max_rows", None, "display.max_columns", None)

input_file = Path.cwd() / 'data' / 'raw' / 'Pandas_Example.xlsx'
df = pd.read_excel(input_file)
data = df.set_index('Week_of_Year')
ar = np.zeros(shape=(52,),dtype=np.float32)
for i in range(len(data)):        
    d=data.index[i]-1
    if data["State_Now"].iloc[i] == "New":
        ar[d]+=1
plt.plot(np.cumsum(ar))
plt.xlabel("week")

Where i have to set the plots exactly in my Code?:

x = range(0,52)
y = ar

plt.plot(x,y)

#19

With just the red code you posted above it’s fine. I cheked it so no need to set x = range(0,52) nor y, matplot will automatically understand plt.plot(np.cumsum(ar)) or plt.plot(ar)
`

#20

@WilfriedF thank you so much for your help, it works now.
Now i can get deeper and deeper into the Options of Analyses :).

Thanks again, Mate!

Simo

EDIT:
One more question came into my mind. Can i somehow change the range of description of the x-axis? I would like to see every single Calendar Week mentioned there (not like now every ten).

1 Like