Posts related to some cool recipes to solve a problem or make your life easier.
View All TagsUsually, training neural networks takes a very long time. Sometimes you need to interrupt the execution of a script when you want to do something else that is equally resource-demanding or turn off the computer. I don't like leaving my PC overnight, so I stop training the neural network and resume training the next day.
I use JavaScript and PHP on a daily basis at work. In these languages there are package managers npm (and more) and composer that allow easy dependency management for each project. So far, I've been writing fairly simple scripts in Python and didn't need any package manager. For upcoming projects, I decided to see what the deal is with dependency management in Python.
This post is a continuation of a previous post about LabelEncoder. This time it will be about a technique called one hot encoding or one-hot. Having categories converted into corresponding numbers, we can also convert them into several columns (the number of columns depends on how many categories there are), which contain zeros and ones, respectively, denoting whether a row belongs to a category or not. We use this method when we use an algorithm that may have a problem with numeric variables (because they assume some order).
Czasami, przetwarzając zbiór danych, mamy do czynienia ze zmiennymi, które są typu tekstowego i przyporządkowują obserwację statystyczną do jakiejś kategorii. Przykładowo, mamy do czynienia z uczniami pewnej szkoły, którzy chodzą do różnych klas (1A, 1B, 1C, 2A, 2B, 2C itd.). Chcemy takie zmienne zamienić na liczby w celu ich dalszego przetwarzania przez jakiś wybrany algorytm np. random forest. Można do tego użyć klasy LabelEncoder z biblioteki scikit-learn.
Sometimes, when processing a data set, we deal with text-type variables that assign a statistical observation to a category. For example, we are dealing with students of a certain school who go to different classes (1A, 1B, 1C, 2A, 2B, 2C, etc.). We want to convert such variables into numbers for further processing by a selected algorithm, e.g. random forest. You can use the LabelEncoder class from the scikit-learn library for this.
As I am still quite a beginner in the world of Python and its libraries, especially those related to machine learning, many things are unknown to me. An example of this is dividing the data set into parts for training and cross validation. So far, I've done it my own way, but why do it this way when we have other tools for it?