My domain is pure data analysis and data science machine learning. The first time I used Weka, five years back, I did a research project. I prefer to work with Weka whenever I have small and clear projects.
Weka is a very nice tool and it helped me to solve any machine learning problem in one minute. In case of machine learning algorithms, classification, or support machines, I used to use this tool to implement those algorithms.
Whenever I get any work on any other platform suppose in hours. So what I initially do, I ran the data set in the Weka platform first. It gives me a clear view that this data set has certain attributes and offers some observations. I can implement different machine learning algorithms if this is a classification.
I use two or three algorithms. If we find that the performance of the logistic regression is good then I can implement those in other platforms also. Weka is a good tool for any analysis.
There are some missing values there. We can replace the missing values using the mean values. I use that filter to see which names were replaced. It's in the filter, then we have to go to that unsupervised, then replace missing values.
I use that filter to replace missing data. Weka has the option to check important attributes. I use that internally, I found that everything is important. Then initially I applied my dataset to implement the classification problem.
There is less demand for projects that require Weka as opposed to R or Python.
Weka is a very nice tool, it needs very small requirements. If I want to implement something in Python, I need a lot of memory and space but Weka is very lightweight. Anyone can implement any kind of algorithm, and we can show the results immediately to the client using the one-page feature. The client always wants to know the story. They want the result.
Accuracy is 90% and we can show the predicted results. What the predicted error is and what the true positive rate is. This is what the user wants. They don't want to know how complex your queries are. They don't bother about it.
It saves time for the developer and provides a summarized view without unnecessary information for the customer.