Problem Solver
Mohamed Elwakdy
Areas Mohamed Elwakdy is Knowledgeable in:
- Signal and Image Processing
- Machine learning
- Data mining
- Machine learning
- Data mining
Techniques Mohamed Elwakdy Uses:
- Applied a preprocessing transform to the input variables in datasets where I rescaled the input variables into the range between 0 and 1.
- Applied the statistical functions to make a good preparation of the data.
- Applied polynomial function and wavelet transform to extract the coefficients in applications related to Pattern Recognition, Speech Recognition and Action Recognition.
- Used many machine learning algorithms such as k-means, Bayesian techniques, support vector machines, neural networks, Adaptive Neuro-Fuzzy Inference System, Decision Tree and Linear Regression to train and evaluate the predictive models.
- Applied the statistical functions to make a good preparation of the data.
- Applied polynomial function and wavelet transform to extract the coefficients in applications related to Pattern Recognition, Speech Recognition and Action Recognition.
- Used many machine learning algorithms such as k-means, Bayesian techniques, support vector machines, neural networks, Adaptive Neuro-Fuzzy Inference System, Decision Tree and Linear Regression to train and evaluate the predictive models.
Mohamed Elwakdy's Problem Solving Skills:
- Computer programming: Matlab, R, Python and C++
- Data mining
- Artificial Intelligence
- Machine Learning
- Microsoft Power BI
Mohamed Elwakdy's Problem Solving Experience:
- -Built models using Machine learning Algorithms such as SVM, KNearest Neighbors, Naïve Bayes (using R/Python) for IRIS dataset to discriminate between different types of flowers (Accuracy 100%) three kinds of flowers: Setosa, Versicolor and Virginica and Kaggle-Titanic data set to predict of died and survived people (Accuracy 98.5%) on aboard the Titanic.
- -Created Power BI dashboards for generating customer insights, share dashboards from the Power BI service and use many interactive visualization tools to represent the data sets.
- -Prediction of medical expenses for the insured population using Linear Regression (using R - Accuracy 75%). The model has the ability to estimate the average medical care expenses for the population segments.
- -Prediction of default risk using Decision Tree Algorithm (using R – Accuracy 74%). Credit approval model is developed to identify factors that make an applicant at higher risk of default. The used dataset includes a large number of past bank loans and whether the loan went into default, as well as information about the applicant.
- -Prediction of cancer disease using K-NN Algorithm (Accuracy 98%) and Linear Regression (using R/Python). K-NN and Linear Regression are used to automate the process of screening for cancer. The model has the ability for early detection of the breast tissue which contains abnormal lumps or masses in. This allowed physicians to spend less time diagnosing and more time treating the disease.
- -Built Speech Recognition Algorithm (SRA) with the ability to discriminate between different samples of speech signals- isolated words- with added background noise (using MATLAB – Accuracy 99%). SRA is very useful for people who have difficulty using their hands which preclude using conventional computer input devices.
- -Discriminated against people walking and running (using MATLAB - Accuracy 99.9%) through the movement of their heads in two dimensions (X, Y). Their ages between 25 to 30 and they are working/running for ~ 10 seconds. The used dataset is Vicon Physical Action Data Set.
- -Built Trajectories Classification Algorithm (TCA) to discriminate between the trajectories of different objects (Tanker Ship and Fishing boat) using MATLAB (Accuracy 99.9%) and C++.
- -Built a Developed Trajectories Classification Algorithm (DTCA) with the ability to discriminate between different objects (using MATLAB): Tanker ship, Fishing boat, Deer, Cattle and Elk through their trajectories (Elk, Deer, and Cattle have similar trajectories (Accuracy 98.83%) and Fishing boat and Tanker ship have non-similar trajectories (Accuracy 99.11%). This work is the first main step to develop an alarm system with the ability to warn road users on crossing animals by specifying the type and place of these animals.
- -Evaluated current strategies relating to data and how it could be used to help executives make business decisions.