109 - Predicting COVID-19 cases using Python

This video explains the process of fitting data to a curve (e.g. exponential) to model data and make predictions. Actively developing COVID-19 data set has been used for demonstration.

References:

[mclark]

Please be consistent with your names. It took me 3 days to find this (Admitedly not trying hard). You previous name was "Analysis of COVID-19 data using Python" and for this one you changed Analysis to Predicting.... Anyway, nice job with your videos but you know.... I finally figured out to change the 108 to 109

[deekshaaggarwal]

Hi Sreeni,


Its a great video. Very much impressed by your teaching capability.


Thankyou for the tutorials.


Kindly update your github repo as I couldn't find the code for 109 - Predicting COVID-19 cases using Python video




Thanks and Regards,
Deeksha Aggarwal
Research Scholar
IIITB, India

[mclark]

I've plotted everything in excel. I think it's possible to do a very good predictive model using these factors: Temperature. Notice that Africa which is hot has very few deaths compared to the wordl. Closeness to the sea: notice that countries around the baltic sea do better (This may just be an anomaly). I don't know much about India, but they also have a very low death rate compared to the west and so does the rest of asia in general. The standard deviation in temperature. The average temperature in the continental US is very warm but with high deviation, so even if the average is 20C, the low may get to 0C and high to 40C whereas and African country may have the same average but low deviation in temperature. I find it hard to understand why the death rates are so different. Every country in the world is trying the best so that's not the reason why a few are seemingly successful, so I think it has to do with temperature. If you put this and also the latitude and longitude of every country into an AI, I think you may be able to get a good model.

[sathvikswaminathan]

The curve is similar to an exponential curve only in the initial stage. This prediction won't work well in the future. A model like SIR or better SEIR is required to predict the future number of cases

[jafarhasan]

@7:25, When Im running
c2 = t, a, b: a*np.exp(b*t), x, y, p0=(20, 0.2))

NameError T
in async-def-wrapper()It shows

NameError: name 'x' is not defined


Although the previous line ran fine where you defined the func using

def plot_exponential_fit_data(d_df, title, delta):

[muhammadalam]

Can you please upload more Deep learning tutorials in the future. All over of your lessons are brilliant!

[Mnikouei]

Hi, Thanks for your nice code and video, it is very very useful. may I ask you to teach me how I can fit the data with SIR model instead of the exponential model?

[babacarlo]

Hi thanks for your video
I have i problem: when a run run the function, it works but i cant see the figure of the matplotlib plot
if you can help

[shaniyakhan]

I cannot print the total cases on y-axis, my graph have it like 0.0, 1.5, 2.0, 2.5 instead of 0, 10000, 20000, 30000, 40000.
Please help me.

[AsifMarazi]

print(f '(y = Ae^(Bx)) A: {A}, B: {B}\n') >>>> syntax error: invalid syntax can you please help