We can never say thank you enough....(+Small Proposal)

F​irst:

I really liked the updated course (the new videos of MobileNet, EffecientNet , Segmentation with U-Net and the new assignments as well as the updated ones: thank you (for all the team) for your work, your dedication,and commitment…really: We can never say thank you enough

S​econd: Small Proposal

If I may: just a small proposal, about the code of plotting the accuracy and loss curves (of the updated Assignment2 (week1):Convolution_model_Application ) : I mean the code provided (at the end).
with pandas library , we can reduce a lot of lines of code (and the curves are “beautiful” and pandas does all the work for us (legend, x_ticks, and so on): only with 7 lines of code (instead of 23) : and it gives the same result (as you see below) :

import pandas as pd
df_loss_acc = pd.DataFrame(history.history)
df_loss= df_loss_acc[[‘loss’,‘val_loss’]]
df_loss.rename(columns={‘loss’:‘train’,‘val_loss’:‘validation’},inplace=True)
df_acc= df_loss_acc[[‘accuracy’,‘val_accuracy’]]
df_acc.rename(columns={‘accuracy’:‘train’,‘val_accuracy’:‘validation’},inplace=True)
df_loss.plot(title=‘Model loss’,figsize=(12,8))
df_acc.plot(title=‘Model Accuracy’,figsize=(12,8));

t​he result (of the code above)


4 Likes

Thank you @Abboura for liking the new content and for appreciating the team’s effort. Indeed a lot of work went into all of this. Really glad you liked it.

And thank you for your suggestions! :clap: I have noted this, I’ll see what I can do about it.

Best,
~Mubsi

1 Like

@mubsi:
This is an example of where a repository of “learner-generated enrichment items” curated by staff and mentors could be of value. It would be nice for people to be (easily) able to find items like this in a few months time.

2 Likes

As I have done with this, I’d want mentors to make Issues of this with Type as suggestion whenever they come across something like this.

Hey @Abboura , thank you again. Your suggestion has been added :wink:

1 Like