I think this question is too vague. Which type of datasets? Which type of analysis did you want to perform? Please provide answers to this questions first and I hope I will be able to give useful answer to your questions.

There are many ways to analyze a dataset. Visualization is one, as using of graphs, charts, etc. allows you to see how your data is in a more understandable manner. The type of visualization differs based on the data you are trying to analyze. You can use Python, SKLearn, T-Sne, and other similar tools and technologies to help you out. I hope this helped in some ways. :) Good luck! Md. Raihan Ahmed

http://dai.fmph.uniba.sk/courses/ICI/References/marsland.machine-learning.2ed.2015.pdf

https://arxiv.org/pdf/1407.7722

I analyzed 3 medical datasets and I successfully predicted the risk with the help of machine learning. Now I didn't want to quit, as deep learning is a part of machine learning I want to expand my analysis with the help of deep learning algorithms.

Toluwase Asubiaro thank you for your answer and you were right my question was vague. Basically I want to analyze medical datasets as well as I want to explore my knowledge about deep learning implementation to computer science/engineering fields because I am from an engineering background and I value your kind opinion.

Francis Jesmar Perez Montalbo Thank you for your answer, from my point of view visualization is must to analyze a dataset. I have already used python, sklearn in my research mentioned above not only that I also found a limitation of using sklearn library of python. Will you please briefly explain where is T-Sne used and for what purposes? Also tell me about the tools(both programming & non-programming) that i should use for prediction?

BERT is a state-of-the art algorithm, BERT is from Google. (BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding) was published barely a year ago and it has been cited more than 3,000 times. Fasttext is another state-of-the-art algorithm, from facebook. These models work excellently for textual data. There are models like Word2vec, sent2vec and doc2vec that are based on Fasttext. Biosent2vec is the biomedical implementation of the Fasttext algorithm, I think it will work better for your biomedical text. Biosent2vec was trained on the whole of PubMed corpus with millions of biomedical publications.

Depending on the type of data, the methods you use may also vary. (Image or signal)

In order to analyze which methods are used in medical data sets, you can look at Survey or Review articles such as https://www.mdpi.com/2414-4088/2/3/47. For this, you can use Matlab, Python and R programming languages.

To answer your question about the deep learning algorithms you can consider. Long short-term memory (LSTM) is a state-of-the-art neural network algorithm for deep learning.

T-Sne is more of a technique, PCA. However, it is very useful for visualizing data with high dimensionality this can be used for some images as well. Images can be turned into tensors and be analyzed further through visualization techniques. There are many more that you could do. This is just some of what I did before. Md. Raihan Ahmed

Please have a look at my dataset.

Cristian Ramos-Vera Thank you for your valuable resources. Actually I was looking for a tool. I want to know is it handy for analyzing Bio-medical datasets. Fulya Akdeniz thank you for sharing such a good article. You also told about R programming language, Can you explain a little bit about getting p-value in R-Studio.

Toluwase Asubiaro Thank you again for suggesting me some article about algorithms like Bert and LSTM. You have pointed the absolute thing that I was working on. Now, I want to know Is there any tool which I can use for Bio-medical datasets.

Francis Jesmar Perez Montalbo Thank you for your kind information. Actually I want to use these algorithms for analyzing Bio-medical datasets. I am providing a picture of my dataset of Ovarian cancer. Just have a look and tell me can I use TSne for analyzing such kinds of datasets?

Md. Raihan Ahmed yes you can my friend. T-Sne is a good tool for dimensionality reduction that could help make your algorithms better by reducing your feature set down to the most significant values. I suggest going over Ensemble learning methods as well to help you with the feature selection process. Then go for a K-Folds cross-validation technique to determine which algorithm works best. Just depends on what you are doing of course. But, more for supervised learning. You can always have me as a collaborator if ever. Good job! Keep it up my friend :)