Machines learn by means of a loss function which reflects how well a specific model performs with the given data. If predictions deviate too much from actual results, loss function would yield a very large value. Gradually, with function, parameters are modified accordingly to reduce the error in prediction. In this article, we will quickly review some common loss functions and their usage in the...
Previously, we train our model using the pre-generated dataset, for example, in the recommender system or recurrent neural network. In this article, we will demonstrate using a generator to produce data on the fly for training a model. Keras Data Generator with Sequence There are a couple of ways to create a data generator. However, Tensorflow Keras provides a base class to fit dataset as a...
Data wrangling (munging), like most data analytics processes, is an iterative one – the practitioner will need to carry out these steps repeatedly in order to produce the results he desires. There are six broad steps to data wrangling, which are: 1. Discovering In this step, the data is to be understood more deeply. Before implementing methods to clean it, you...
This post will demonstrate onehot encoding for a rating matrix, such as movie lens dataset. One-hot encoding Previously, we introduced a quick note for one-hot encoding. It is a representation of categorical variables as binary vectors. It is a group of bits among which the legal combinations of values are only those with a single high (1) bit and all the others low (0) Rating matrix If you are...
We will build a recommender system which recommends top n items for a user using the matrix factorization technique- one of the three most popular used recommender systems. matrix factorization Suppose we have a rating matrix of m users and n items. The rating of user to item is . Similar to PCA, matrix factorization (MF) technique attempts to decompose a (very) large matrix () to smaller...