Real State Investment

There are many variables involved in real estate financial decisions and more than anything when evaluating whether a potential property is a good investment or not. Data has made this process much easier for the decision-makers as it has significantly reduced the risks and uncertainties that are associated with real estate investing by comparing markets and properties in a much efficient way. For this project, the Client wanted to evaluate the roofstock website portfolio in a way that could give him a good insight in which areas or state was worth to invest, i.e. best price and best cap rate.

Methodology:

• Automated Data Extraction from roofstock website, by coding a script in python using selenium library,
• Data Cleaning and Preparation: preprocessing the data to prepare it and transforming it into an appropriate format. Used of Geopy library to get the latitude and longitude of the properties,
• Data Visualization: built a dashboard in Power Bi, showing the relationship between the locations, prices, and cap rate,
• Built a clustering model using DBSCAN algorithm based on Prices and Cap Rates.

Cluster 0, Avg Cap Rate: 12.69 | Cluster 1, Avg Cap Rate: 8.70 | Cluster 2, Avg Cap Rate: 10.84 | Cluster 3, Avg Cap Rate: 11.88 Cluster 4, Avg Cap Rate: 9.74 | Cluster 5, Avg Cap Rate: 8.19 | Cluster 6, Avg Cap Rate: 9.71

Predective Analysis
By creating simple running decision making processes with predictive analytics, investors will be able to produce new insights and impartially estimate when and where to invest, moreover optimize their selections and determine the real value of properties. For this case, I created a model which could predict the capitalization rate of the property on sale. Cap rate is one of the most popular measures and it indicates the rate of return that is expected to be generated on a real estate investment property. More details

Close Project

Energy Efficiency-Room Occupancy Detection

Using statisticals classification models to predict occupancy in an office room, is a considerably great strategy for energy saving and improvement of energy efficiency. There have been studies that support that Heating Ventilation and Cooling(HVAC) represents one of the largest sources of energy consumptions in residentials and buildings. However, by implementing smart monitoring it is reported to save energy by 30 to 42%. This data set has been sourced from the Machine Learning Repository of the University of California. This project supports some of the results by evaluating the datasets as part of the article "Accurate occupancy detection of an office room from light, temperature, humidity and CO2 measurements using statistical learning models". My goal was to evaluate different machine learning techniques and their accuracy.

Methodology:

• Selection and Understanding of Training dataset,
• Data Cleansing: Elimination of duplicated values. Dates are usually useful as continuous values to evaluate how all the physical factors behave through time,
• Exploratory Data Analysis: Visualization and Evaluation of Correlation between physical factors,
• Loading of Test dataset and preparation for the model,
• Training of the Model: applied of machine learning classification model,
• Evaluation of accuracy and comparison between the predictive methods,

CO2 and Light have a good correlation with Occupancy level. Indeed levels CO2 increase when people are in a room due to the process of breathing. Why this is important? Since we are trying to determine the probabilities of occupancy in a room, these factors improve the predictive model through a classification approach. Whenever a sensor determines a decrease in the CO2 level and lights, it could be decisive to tell the room is not being occupied, improving therefore, energy saving from HVAC system. More details

Close Project

Panama Metro Stations Analysis

The development of the Panama Metro system is a long-term process, and it carries along with it ramifications of positive impacts, that could be analyzed in order to maximize the opportunities. My purpose was to get an insight by analyzing a ratio of venues that characterize the transit of people around each metro station, I could expect as a result, a classification of the main commercial activities in its surroundings and primary usage of each station. The objective is summarized as the following:

• Find places for new business development,
• Identify the urban structured to raise more strategies for the urban planification’s institutes,
• Relate the travel demand characteristic in each place,
• Boost the tourism activity by recognizing the principal venues in each station.

Methodology:

• Data Cleaning and Transformation: dropped redundant data and prepared for statistical analysis,
• Data Acquisition: got a list of stations and their coordinates, Connected to Foursquare API and got data of the venues, scraped data of Foursquare Main Categories from the website,
• Exploratory Data Analysis: compared the main categories that characterize the venues surrounding the Stations,
• Applied K means Machine Learning Algorithm,
• Got final insights represented in bar charts and shared results.

More details

Close Project