Disaster causes life in danger for all types of animals. Today due to climate changedisaster occurring frequently. Moreover, when people traveling at sea on a largeship (e.g., The Titanic), the disaster can lead to a significant death of passengers.However, some passengers save their lives in the Titanic accident. It means thatthere are still some possibilities for passengers to be alive. This project aims topredict which passengers have more possibilities to be alive among all passengers.For this purpose, we believe there are features that can be used to design learningapproaches to predict passenger life, such as seat position and gender.More precisely, we use the Titanic data set to execute our learning model to predictpassenger safety. The experimental result shows the effectiveness of our model.
The motivation of this project is mainly to help people make a decision about theposition of the seat that guarantees the possibility of surviving when the shipwreckfalls into a disaster. With this motivation, we choose a dataset (e.g., Titanic dataset)to analyze which type of passengers has the most chance of surviving. For doing this,we try to find out the answers of the following questions: Which type of passengers(e.g., male, female) have great chance to survive? Which part of a ship sunk later?etc.We try to figure out above the questions. And we are trying to find a better solutionfor the questions answer, through designing a learning model. That is our main focusand motivation to build this model.
We take a dataset from kaggle.com. Nowadays Kaggle arrange data analysis com-petition. We took a challenge to solve this problem. That was a competition forbuildup a model that can give better accuracy for predicting the survival rate. Weuse here basic machine methodology, some famous algorithms. Our main focus is toincrease the accuracy level. Here we try to learn basic machine learning methodology,some famous machine-learning algorithms. Besides this, we apply all methodologiesand algorithms on the dataset to find a better accuracy rate.
Here we use a well-known machine learning algorithm to find out a bettermodel that can give us our expected value. Better accuracy percentage For
Support Vector Machine (SVM): 73.3%
Decision Tree (DT): 74.19%
Random Forest (RF): 80.7%
In this paper, we presented an effective partially supervised methodology forpredicting a man who survived or not. We used DT (Decision tree) andRF (Random forest) machine-learning algorithm to predict our survival rate.These two supervised machine learning algorithms showing high performancewith label data.We introduce a sample learning-based method for predicting our survivedlabel. Our objective is to predict the survival yes or not. By training onlynormal data without feature engineering this survival accuracy is under 50percent After feature engineering survival accuracy is above 70 percent. Inthe future, we will try to increase our accuracy rate of about 100 percent. Boththe qualitative and quantitative results on challenging datasets show that ourmethod outperforms well.