jainmansi/BayAreaBikeShareAnalysis

A Hadoop Mapreduce project to analysis SF Bay Area Bike Share Dataset.

Introduction

In this project, I am analyzing SF Bay Area Bike Share dataset. To do so, I am using following datasets:

1. Trips Dataset
2. Station Dataset
3. Weather Dataset
4. Yelp Customer Reviews Dataset

Through this analysis, I aim to achieve a better understanding of how different factors affect the bike sharing and how BABS has been performing over the course of time.

Click here to go to the dataset.

Technology Stack

1. Hadoop MapReduce
2. HBase
3. Hive
4. Pig
5. Weka
6. Tableau

Analysis results

Note: All the charts are plotted using Tableau

Analysis 1: Binning pattern to determine distribution of rides in a day

Using this analysis, I aimed at finding the spread of rides over the hours of the day.

Analysis 2: Subscriber vs. Customer during each year

In this analysis, I have used custom writable objects to store the running total of the count for each kind of user. The result obtained can be understood from the following tableau chart:

Analysis 3: Distribution of rides over months

To determine the distribution of rides over the year for each month, I have used counting with counters pattern.

Analysis 4: Determining Top 6 stations based on the number of rides

Here, we are employing Top K values pattern to determine top 6 most busiest stations in the bay area. These are the stations from where most number of rides start.

Analysis 5: Distribution of rides based on temperature

In this analysis, we are determining how the mean temperature of a day affects the turn up of people for bike share. We are making this analysis for top 5 busiest stations.
To achieve this, we are join two datasets using inner join. We are also employing secondary sorting, inner join, top k pattern and chaining techniques.

Analysis 6: Sentiment Analysis of user's yelp reviews

I have used Yelp API to fetch data of customer’s review for BABS service. A very naïve method is used to analyze sentiments of users.

Analysis 7: Using Pig to find duration of trip for start station, total trips started and average trip duration

I have used Pig in local mode for this analysis. Following are the commands

1. Start Pig

./pig -x local

2. Load table data

tripdata = LOAD '/Users/mansijain/Desktop/BABS-Dataset/trips/trips.csv' USING PigStorage(',') as (trip_id: int, start_date: chararray, start_station: chararray, start_term: chararray, end_date: chararray, end_station: chararray, end_term: int, bike, zip_code: int);

3. Group data by start station column

split_station = GROUP tripdata BY start_station;

4. Perform aggregate function

results = FOREACH split_station GENERATE GROUP AS start_station, SUM(tripdata.duration), AVG(tripdata.duration), count(trip_id);

5. Store results in local fs

STORE results INTO '/Users/mansijain/Desktop/BABS-Dataset/results/start_stn' USING PigStorage(',');

Following chart was obtained:

Analysis 8: Using Pig to find duration of trip for end station, total trips started and average trip duration

A similar pig query is used to determine data associated with the end station.

1. Group data

split_station = GROUP tripdata BY end_station;

2. Aggregate data

results = FOREACH split_station GENERATE GROUP AS end_station, SUM(tripdata.duration), AVG(tripdata.duration), count(trip_id);

3. Store data

STORE results INTO '/Users/mansijain/Desktop/BABS-Dataset/results/end_stn' USING PigStorage(',');

Analysis 9: Using Hive to analyze total stations and dock counts in all the cities of Bay Area

I have used Hive for this analysis. Following are the steps I followed:

1. Create table

CREATE TABLE stationTable (id int, name string, lat double, long double, dock_count int, landmark string, date_inst date) row format delimited fields terminated by ',' STORED AS TEXTFILE;

2. show tables

show tables;

3. Insert vales

INSERT OVERWRITE local directory '/Users/mansijain/Desktop/BABS-Dataset/hive' ROW FORMAT DELIMITED FIELDS TERMINATED BY ',' SELECT landmark, count(id), sum(dock_count) FROM stationTable GROUP BY landmark;

Analysis 10: Analyze the spread of all the stations in the cities of Bay Area

Using distinct filtering pattern, I analyzed all the stations present in the bay area. I have utilized latitude and longitude column to visually see the locations on a geomap.

Analysis 11: Analysis of trips based on start and end station

The following analysis determines the sum of all the rides between two stations. I used composite keys to determine these values.