If presented with two customers with their respective profiles, how do you determine who will get a loan? The goal of this project was to build a model that lenders can use to help make the best financial decisions regarding the customers to lend and those not to lend. Therefore, the main goal is to build a state-of-the-art credit scoring model by predicting the probability that somebody will experience financial distress in the next two years. This is a binary classification problem with classes; 0 : Not deliquent & 1: Deliquency (Kaggle Challenge).

Algorithms : Logistic Regression, Random Forest, Decision Trees and k-Nearest Neighbors

How can the task of face detection be achieved? On this topic, earlier studies have proposed a framework for face detection based on multi-task cascaded CNNs. Experimental results and my other proposed framework show that my MTCNN techniques consistently outperform most of the main techniques. The second experiment is likewise accomplished using the Haar cascade set of rules on my dataset.

Algorithms : Convolutional Neural Network, MTCNN, Haar Cascade

This study examines the impact of introduction of futures trading on the spot price volatility in the commodity market. The paper considers the United States of America, South Africa and Ethiopian economies. Three commodities i.e. coffee, maize and wheat from New York Mercantile Exchange, South African Futures Exchange and Ethiopian Commodity Exchange are analyzed. ARCH LM test is used to check for heteroskedasticity and GARCH and EGARCH are used to check for the behavior of volatility for the pre- and post-futures periods. This paper has focused on the overlooked factor by earlier researchers, i.e. of economic-gap amongst countries, in looking at the impact of the futures trading on the spot price variation.

Keywords: derivatives, futures exchange, agricultural commodities, spot price volatility

With the advent of high frequency data, the sum of squared returns between trades which is the most common estimator, is biased by microstructure effects like bid-ask bounce thus the need to drop most of the data. Nonetheless, a number of alternative estimators that make efficient use of the available data have been developed. However, choosing an estimator is not trivial since the study of their relative merits focuses on the speed of convergence to their asymptotic distributions. The paper is an effort towards estimating a covariance matrix using high-frequency data (quadratic covariation) from the portfolio selection perspective. Covariance matrices based on intraday returns were constructed and evaluated.

Keywords: High Frequency Data, QMLE, Portfolio Optimization, Covariance Matrix

The study was conducted across three agricultural commodities, namely; corn, cotton and wheat against a number of weather variables, including; temperature, rainfall, precipitation, sea level pressure, snow, fog and hail. This is after dropping a number of weather variables that are highly correlated from the correlation matrix. The results indicate significance of these exogenous variables in explaining the price variations as seen from the p-values of 0.00 across the three commodities. The SARIMAX prediction also did well for Corn and Cotton prices as seen by the sufficiently high r2 score as opposed to the estimate gotten from wheat. (Yet to study the same applying AI techniques).

Keywords: SARIMAX, Weather Shocks, Price Volatility

Energy remains an essential pillar for human livelihood for its necessity in sustainability and development of any current civilization. One of the main feature distinguishing it from most of the other commodities is the mean-reverting characteristic together with evident spikes and high volatility as seen for electricity and crude oil prices. A number of works have employed the Ornstein–Uhlenbeck process to model directly the dynamics over time of different commodity spot prices under reduced-form one-factor models (Ribeiro and Hodges, 2004). Nonetheless, in recent studies, crude oil spot price has also been modelled as a jump-diffusion process, as attributed to Merton (1976), as in Jorion (1988) and Ball & Torous (1983).

Keywords: Crude Oil Price Modeling, Jump-diffusion models, Poisson process

Energy remains an essential pillar for human livelihood for its necessity in sustainability and development of any current civilization. Here, we derive the analytical solution to the stochastic differential equation for the Ornstein-Uhlenbeck process: dXt=κ(θ−Xt)dt+σdWt where the Wt is a standard wiener process, and κ>0, θ and σ>0 Motivated by the observation that θ is supposed to be the long-term mean of the process Xt, we can simplify the above SDE by introducing the change of variable; Yt=Xt−θ that subtracts off the mean. The Yt satisfies the SDE; dYt=dXt=−κYtdt+σdWt for this SDE, the process Yt is seen to have a drift towards zero value, at an exponential rate κ. This motivates the change of variables; Yt=e−κtZt ⇔ Zt=eκtYt, which should remove the drift.

Keywords: O-U process, Carbon emission

This project is an attempt at forecasting the stock prices for Apple (AAPL), Microsoft (MSFT) and Meta (FB) with the use of the Long Short-Term Memory (LSTM) neural network. The data was extracted from Yahoo finance for daily, weekly and monthly intervals’. Apple and Microsoft stock prices covered the period from 01-01-1999 to 12-08-2022 while the Meta data available covers from 19-05-2012 to 12-08-2022. The forecasting was done with use of the closing stock prices for the respective time frequency. The testing result confirm that our model is capable of tracing the evolution of opening prices for both assets. Future work on the topic could try more range of hyperparameter tuning (number of epochs and batch sizes) to get to the global minima of the loss function and check for any improbvements in the model accuracy.

Keywords: Time Series, LSTM, RNN