Bachelor-Thesis

KO | EN

gitlite — search

#javascript #python #hacktoberfest #react #ai #typescript #llm #go #golang #android #machine-learning #rust #deep-learning #linux

Bachelor-Thesis

(★ 9)

Deep Learning Algorithms Applied to Blockchain-Based Financial Time Series

Bachelor-Thesis 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Bachelor Thesis
Deep Learning Algorithms Applied to Blockchain-Based Financial Time Series (FTS).

## 1. Exercises 
The main project was approached via multiple sub-tasks or exercises, before building up the final model. All R scripts corresponding to each sub-task can be found in the `src` directory, with corresponding datasets in the `datasets` directory. A short PDF document accompanies each exercise to present a summary of results, all of which can be found in the `doc` directory.

### Exercise 1: AR Yule-Walker, AR Burg
This exercise investigates linear AR Yule-Walker and AR Burg models applied to FTS prediction.

### Exercise 2: ARMA & ARIMA
These exercises investigate linear ARMA & ARIMA models applied to FTS prediction and why linear models are not necessarily the most suited for modelling FTS.

### Exercise 3: ARCH & GARCH
This exercise investigates linear ARCH & GARCH models applied to FTS prediction and why linear models are not necessarily the most suited for modelling FTS.

### Exercise 4: SLP & NN
This exercise investigates non-linear Single Layer Perceptron (SLP) & basic Neural Networks (NN) and their performance for modelling FTS.

### Exercise 5: NNetAR 
This exercise investigates Single Hidden Layer Neural Networks (NNetAR algorithm) and their performance for modelling FTS.

### Exercise 6: SOM (Animals)
This exercise investigates Self Organising Maps (SOM) and their performance in unsupervised learning.

### Exercise 7: Forex vs. BTC 
This exercise investigates the differences in applying the AR, ARMA, ARIMA, ARCH, GARCH & SLP models to Forex & BTC FTS (binary only).

### Exercise 8: NARMAX
This exercise investigates applying the Nonlinear Auto Regressive Moving Average model with eXogenous inputs (NARMAX) to both predicting Forex & BTC FTS (trend & binary).

### Exercise 9: SOM (BTC/Forex)
This exercise investigates Self Organising Maps (SOM) and their performance in unsupervised learning applied to Forex & BTC FTS.

### Exercise 10: ESN (BTC/Forex)
This exercise investigates Echo State Networks (ESN) and their performance in unsupervised learning applied to Forex & BTC FTS.

### Exercise 11: LF Granger Causality (Cryptocurrencies)
This exercise investigates Low Frequency Granger Causality in cryptocurrencies.

### Exercise 12: HF Granger Causality (Cryptocurrencies)
This exercise investigates High Frequency Granger Causality in cryptocurrencies.

### Exercise 13: HF NN (BTC)
This exercise investigates a preliminary High Frequency NN for forecasting BTC close price with OCLH data.

### Exercise 14: HF NN (DASH)
This exercise investigates a slightly improved High Frequency NN for forecasting DASH close price with delayed time series and exogenous BTC delayed inputs.

### Exercise 15: Volatility (BTC/Forex)
This exercise investigates briefly fully Bayesian estimations of stochastic volatility via Markov chain Monte Carlo methods in the BTC-USD pair compared to other typical Forex currencies. For context only.

### Exercise 16: Further NN Investigations (Cryptocurrencies)
This exercises looks at the impact of going from HF minute-data to MF-hour data as well as other simple NN architectures.

### Exercise 17: Deep NN Investigations (Cryptocurrencies)
This exercises looks at Deep NN architectures using Keras, TensorFlow & CUDA.

### Exercise 18: Hour Granger Causality (Cryptocurrencies)
This exercise investigates Hourly Granger Causality in cryptocurrencies.

### Exercise 19: Further DNN Investigations (Cryptocurrencies)
This exercises looks at further DNN architectures (up to 20x20 with 2 hidden layers) using Keras, TensorFlow & CUDA.

### Exercise 20: Live Algorithmic Trading (Cryptocurrencies)
This proof of concept exercise uses the most performant algorithm developped thus far to trade live on the Poloniex exchange.

## 2. Mathematical Fundamentals 
These equations cover the bare fundamentals of all the models used. This can also be found in the `doc` directory.

## 3. Project Progress report
(Complete, graded 80.50%)

## 4. Final Bachelor Thesis
(Complete, graded 92.00% overall)

Bachelor-Thesis

Bachelor-Thesis 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Bachelor-Thesis

Bachelor-Thesis 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README