Decision Tree in Python | Machine Learning Algorithms | Dr. Ernest P. Chan | Quantra by QuantInsti

** Decision Tree for Trading: https://quantra.quantinsti.com/course/decision-trees-analysis-trading-ernest-chan ** FREE PREVIEW!

Timestamp:
00:13 - 00:37 - Decision Tree algorithms
00:44 - 00:52 - Aim of decision tree
00:53 - 01:11 - Nodes types
01:12 - 02:25 - Structure of simple decision tree
02:26 - 03:08 - Decision tree rules


In this video, we will learn about decision trees which are widely used in machine learning. Decision tree algorithms fall under the category of supervised machine learning and are used for both classification and regression problems. We have two main types of decision trees. Classification trees in which the target variable is categorical, i.e. of Yes and No types and Regression trees in which the target variable is of continuous data types as a number. Decision trees are so-called because they use a tree-like model to visually and clearly represent decisions and decision making. The aim of a decision tree is to arrive at possible outcomes for the target variable which is based on the choices of predictor variables. A decision tree starts from the root node which does not have any incoming pathways. A node with an outgoing pathway is called an internal node. Each internal node again splits the predictor’s space into two or more sub-spaces. The remaining nodes are called leaf nodes or decision nodes. Let us understand the structure of a simple decision tree. The example dataset comprises four predictors namely Past Trend, Open Interest, Volume, and Gap Price for a stock. The target variable Return indicates the daily change in the stock price which can be either up or down. In this example, Trading Volume is the root node that gets selected from all the available predictors based on a criterion like Gini index or Information gain. When Trading Volume is High the dataset before the split takes place would look as shown here. Out of the available predictors, Open Interest is selected as an internal node based on the set criteria and it further splits the dataset into smaller datasets. After the split, where Open Interest is low, the remaining dataset would look as shown here. This splitting process is carried out along all the paths of the decision tree. When a decision tree is completely constructed, each leaf node is assigned to one target class. In case of regression trees, the leaf node may hold an average value of the target class. A decision tree can be transformed into a set of rules. For example, as seen in rule number one if the trading volume is high and the open interest is low then the predicted return will belong to the “Up” class. If the trading volume is high and open interest is also high then the predicted return will belong to the “Down” class. If the trading volume is low and the past trend is negative then the predicted return will belong to the “Down” class. Finally, If trading volume is low and the past trend is positive then the predicted return will belong to the “Up” class. Thus, decision rules are formed by mapping from the root node of the decision tree to the leaf nodes one-by-one. This concludes the video lesson. In this video, we understood the basic structure of decision trees, their construction and how decision rules are formed. In the next video, we will cover decision tree inducers and other aspects of decision trees.

Quantra is an online education portal that specializes in Algorithmic and Quantitative trading. Quantra offers various bite-sized, self-paced and interactive courses that are perfect for busy professionals, seeking implementable knowledge in this domain.

Find more info on - https://quantra.quantinsti.com/
Like us on Facebook: https://www.facebook.com/goquantra/
Follow us on Twitter: https://twitter.com/GoQuantra