I want to use Thyroid data-set(UCI repository) for my machine learning approach. The data-set having 21 attributes and 7200 instances. 15 instances out of 21 are binary and rest are continuous. Please suggest me how to pre-process this data so that i can use it for my work.
Max-min normalization can be used for contiguos features and no need to pre-process the binary features. I am not very sure, but i think it could give good result. You can refer to this page. I wish it will help you.
http://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.MinMaxScaler.html
depends on your data.. however normalization and feature selection can be enough
Various steps can be performed for pre-processing dataset based upon your data and your research problem. The most common steps are: 1)exploring data in the initial stage by plotting it to understand the nature of your data. You can plot the correlations in data 2) Describe your data like finding mean, mode, median and range to see the distribution of data. different visualization(histogram, line graph, box plot) can help you in this step. 3). After doing the first two steps, you should be able to explore and understand your data, now remove inconsistent values, duplicate records, missing values, invalid data and outliers etc to prepare your data for applying machine learning models. You can also perform normalization, feature selection, dimensionality reduction etc if required.
- Badges
- Science topic