I have 8 years of daily data. I want to graph all values per day of the week, per weeks in a year and per month in a year. How do I do that?

user_10868993 0
Viewed 2065

I want to be able to visualize my data points per days of the week, per weeks in a year and per months. I was able to visualize my data per year. But when I adjust the code for Monthly and weekly, the x-axis remains as per year.

I have 8 years of hospital records. My data is organized into 2 columns. Column number 1 is my "dates" column starting from 2010-03-10 and ending at 2017-12-31. Column number 2 is my value column. This column lists if I've had a patient come in for treatment or not. The values in column 2 are 0 or x numbers. For example 0 meaning I've had no patients x meaning I've had x number of patients. When I try to graphically represent this data, it only counts the largest x number I've had per week.

df=pd.read_csv('Data 3.csv', parse_dates=["Dates"], index_col="Dates")
# create the plot space upon which to plot the data
fig, ax = plt.subplots(figsize = (10,10))

# add the x-axis and the y-axis to the plot
ax.plot(df.resample('Y').sum()['Total # Events'],color = 'blue')

# rotate tick labels
plt.setp(ax.get_xticklabels(), rotation=45)

# set title and labels for axes
ax.set(xlabel="Years",
ylabel="Total # of Events",
title="Yearly Treatment Events from 2010-2017");

Graph result

So I get the correct graphical figure. But when I change the (df.resample('Y').sum() to ('M') from ('Y') for monthly I get a graph that displays a yearly X-axis and values. How can I change this to get monthly X-axis and Weekly X-axis?

pandas data-visualization matplotlib python
2 Answers

So, as I had assumed, this was easy given the dates were available.

So first create new columns denoting which week, year and month a date belong to. For that you need to set Dates as index and parse_dates while reading the csv:

data=pd.read_csv('Data 3.csv',index_col='Dates',parse_dates=True)

Now you can create the three columns:

data['Week']=data.index.week
data['Month']=data.index.month
data['Year']=data.index.year

This will give you data like:

            Total # Events  Week  Year  Month
Dates                                        
2010-03-10               0    10  2010      3
2010-03-11               4    10  2010      3
2010-03-12               0    10  2010      3
2010-03-13               0    10  2010      3
2010-03-14               0    10  2010      3
2010-03-15               0    11  2010      3
2010-03-16               0    11  2010      3
2010-03-17               0    11  2010      3
2010-03-18               2    11  2010      3
2010-03-19               0    11  2010      3

Rather than sharing whole code for you, I'll just show the Year output and how it works, rest I hope you can do by yourself, or you won't learn anything that way.

So the next step is:

Yearwise=data.groupby(by=('Year')).sum()['Total # Events']

And that's it, plot it:

plt.figure(figsize=(14,10))
Yearwise.plot()

And the graph as you have shared too, is: enter image description here

Remember, for Month wise, you'll need Year and Month together for grouping or it will end up grouping all the same numbered months together. What I mean is something like this- 

by=('Year','Month'))

Rest I am sure you can figure it out on your own. Still if you're getting error, let me know.

user_21263570

Following the guide on matplotlib.org entitled "Date tick labels", I suggest playing around with these parameters:

# format the ticks
ax.xaxis.set_major_locator(years)
ax.xaxis.set_major_formatter(yearsFmt)
ax.xaxis.set_minor_locator(months)

Try for example adding only this line to your code after ax.plot():

ax.xaxis.set_major_locator(months)

Bear in mind that although you are resampling your data, it is not changing the start or end dates. Your graph will look very similar except for it will now be smoother; it will not have 7 points but 7 * 12 points.

user_18382570
Related