Study of the most relevant companies in the technology sector and classification using historical data

BigData2020MasterInformatica / FinanceAnalysis

MIT License

0 stars 0 forks source link

Study of the most relevant companies in the technology sector and classification using historical data #21

Closed FernandoGD97 closed 3 years ago

FernandoGD97 commented 3 years ago

The next development of the project will be divided into several parts:

Data collection and processing: We will get through the API Financial Modeling the history of several leading companies in the technology sector.
Elaboration of a classification model: From the above data, we will develop a model to predict the closing price.
Visualization and analysis of the results obtained: Finally, using the results gotten we will represent to study their behavior and to be able to obtain conclusions.

DRTorresRuiz commented 3 years ago

Hi @FernandoGD97 @benjivega @mchuckles, I'd like to integrate #19 #22 with this issue. How do you think we can do it? Should we pass an additional argument to the to_buy_or_not_to_buy() function defined in #22 to allow the use of different classifiers? Since we do not have a lot of extra time, what classifiers do you think we should include?

FernandoGD97 commented 3 years ago

Hi @DRTorresRuiz . Seeing that we are short of time, I would recommend that you indicate that you are working with time series and we with classification algorithms in this iteration. Implementing different notebooks. Anyway, the classifiers we have used are:

Linear Regression
Gradient Boosting Regressor
Isotonic Regressor

DRTorresRuiz commented 3 years ago

Hi @DRTorresRuiz . Seeing that we are short of time, I would recommend that you indicate that you are working with time series and we with classification algorithms in this iteration. Implementing different notebooks. Anyway, the classifiers we have used are:

Linear Regression

Gradient Boosting Regressor

Isotonic Regressor

Thank you for answering! I'll check these algorithms and try to integrate them with #19 and #22, they are not classifiers tho.

DRTorresRuiz commented 3 years ago

I have included the following classifiers in the function to_buy_or_not_to_buy() in #24 so far:

Multilayer Perceptron
Random Forest
Decision Tree
~~Gradient-Boosted Tree~~

It exits with an error when using Gradient-Boosted Tree: Labels must be in {0,1}; note that GBTClassifier currently only supports binary classification.

This is because we use 0, 1 and 2 labels to classify.

~~LinearSVC~~

LinearSVC only supports binary classification. 3 classes detected in LinearSVC

I have just modified the Wiki with the previous mentioned changes.