How can we find the optimmal number of Eps and min_samples? Is there any method similar to elbow method under Kmeans by which we can find the best values?
I have a similar query how do I choose values for eps and min_samples for different data sets, a few data sets have huge values and a few have a smaller values. Thanks in advance if anyone answers the same
why is the elbow method used here to set the value of Epsilon? Epsilon is the radius of the circles in DBSCAN right and elbow method gives us optimal number of clusters for K-means? how are these two related exactly?
I know this is kind of old now, but I think it was supposed to be used to make sure you get the optimal number of clusters. He had clusters 0-4 which is 5 clusters, and that matches up with the result of the elbow method. I used all the features of the data (not just 2), so I had to guess-and-check values of epsilon and minimum samples until I got 5 clusters.
Because we fit the data on the training set only because we use training data for training the model and predict on the test data because that data is not trained by the model it is unseen data.
Please turn on the automatic subtitles/captions, it helps me understand better. I'm a visual learner. I'm also really bad at understanding accents. Also the playlists are gone.
Trust me bro. You have done great job. Just amazing. Very helpful and simple code. Thanks and keep it up.
Thank you Pankaj for your kind words.
How can we find the optimmal number of Eps and min_samples? Is there any method similar to elbow method under Kmeans by which we can find the best values?
I will get back on this soon.
I have a similar query how do I choose values for eps and min_samples for different data sets, a few data sets have huge values and a few have a smaller values. Thanks in advance if anyone answers the same
why is the elbow method used here to set the value of Epsilon? Epsilon is the radius of the circles in DBSCAN right and elbow method gives us optimal number of clusters for K-means? how are these two related exactly?
I know this is kind of old now, but I think it was supposed to be used to make sure you get the optimal number of clusters. He had clusters 0-4 which is 5 clusters, and that matches up with the result of the elbow method. I used all the features of the data (not just 2), so I had to guess-and-check values of epsilon and minimum samples until I got 5 clusters.
Where can I find this jupyter notebook of yours
Please find the dataset and jupyternotebook link: github.com/siddiquiamir/Python-Clustering-Tutorials
Man do you know where can I find the code of DBSCAN implemented from scratch but for multidimensional datasets? Not only 2 x's but x1,x2,x3,and x4
I am not sure but I will have to look for it.
why do we use fit_predict and not fit on the train set and then predict on the test set? thanks!
Because we fit the data on the training set only because we use training data for training the model and predict on the test data because that data is not trained by the model it is unseen data.
Thanks brother ! great video
Can I have the jupyter notebook ?
Sure, please find the link for notebook and dataset: github.com/siddiquiamir/Python-Clustering-Tutorials
@@StatsWire Thanks a lit man
@@suridtheanalyst6770 You're welcome
Thank you for the video, replace all that plotting mess with this line :
sns.scatterplot(x="x_axis", y="y_axis", hue="labels", data=df)
Keep it up !
Thank you for providing the shortest code:) I hope this will help others
Could you change the description of the video and insert there a link to the code and to the data set, please?
Thanks for the suggestion. I will do it right away.
Thank you so much sir!
You're welcome!
After I click "Run" at 1:39, I get a "NameError: name 'df' is not defined". I don't know why it is different for you
Hi please see if df is saved or not. Before running that line use print(df) and see if you are getting error or it is printing df
Can u give the dataset in the description box?
Please find the dataset and jupyternotebook on my github account: github.com/siddiquiamir/Python-Clustering-Tutorials
Very helpful video
Thank you
How to detect anamoly using dbscan can u provide me the code
That's a different topic. I will have to make a separate video on that
@@StatsWire thanks
Please turn on the automatic subtitles/captions, it helps me understand better. I'm a visual learner. I'm also really bad at understanding accents. Also the playlists are gone.
Hi, I am not aware if I can still do it. Is it possible to do it now? I will do it right away if that is the case
how to plot a single cluster I got only one cluster
That's not correct. Because if there is only one group then why there is a need to do clustering.
How to calculate the silhouette score ?
You can refer to the official documentation: scikit-learn.org/stable/modules/generated/sklearn.metrics.silhouette_score.html
Can u help me? I have a Doubt
Yes, let me know
I am doing my thesis using DBSCAN
Can u help me to do?
Can I contact u?
@@fathimafarha8217 you can connect on Instagram stats_wire
Nice tutorial
Thank you
my code is not reading the dataset
please help
What problem are you facing?