Impact of a Non-Dedicated U-Turn on Traffic

Non-dedicated U turn has a direct effect on road safety, capacity and congestion during the traffic flow. U turn can have significant supremacy on traffic flow and headway. Therefore to study the impact of non-dedicated u turns on traffic is the ultimate requirement of the current time. This is a microscopic traffic study in which the data from a U turn (33°59’48.2"N 71°27’30.2"E) on road leading to Hayatabad and Karkhano in Peshawar is evaluated in terms of headway, speed and flow rate of traffic. Factual data is presented which shows that the average time headway surges when the traffic is interfered by the U turning vehicles. The probability density functions and cumulative density functions fit to the datasets of headway are then evaluated by the techniques of anova analysis to determine which distribution is the most suitable one for the data. Distribution data specific with the interfering U turn was taken in a separate set and evaluated. The result obtained show that the Burr Distribution and Generalized Extreme Value Distribution are the optimum to illustrate the headway data of traffic being interfered by U turning vehicles. This ligitimize the utilization of various time headway distributions of vehicles being interfered by U turning for traffic modeling.

Location Based Sentiment Mapping of Topics Detected in Social Media

Social media has risen as a powerful social link among people. It gives them an easy, unsupervised, rapid and flexible platform for sharing information with each other. It serves like a virtual environment where there is no direct or indirect executive power to dictate their opinions. They feel free to appreciate or criticize the actions of others. That is why social media posts carry true reflection of public feelings over any social event. This very nature of social media content makes it a valuable asset for research community to dig deeper and find meaningful concerns out of its huge chunks. This research work explores how people across different regions of Pakistan think about any social media topic. The research work proposed a model to extract spatial, topical and sentimental information of social media posts and combine them together to perform location-based sentiment mapping of the topics detected in social media. The study area has been restricted to Pakistan.The research work started with collection of tweets using streaming API of the Twitter. The information contained in a 3rd party dataset worldCities was used to filter out Pakistani tweets and perform indirect conversion of user profile location into geo-location, as geo-location was directly not available with tweets. Probabilistic topic model; Latent Dirichlet Allocation (LDA) was used to extract latent topics from the dataset. CRAN library was used for sentiment classification of the tweets into 3 classes: positive, negative and neutral. The processed dataset was uploaded to a persistent storage Google Drive and then imported into a web based interactive self-written Shiny Application to read geo-spatial, topical and sentimental information of the tweets from processed dataset and produce interactive geographical map. The interactive geographical map has also been supported by heatmap, scatter plots, bar graphs and pie charts in order to enhance visualization process. The novelty of this research work is its location-based visualization of the strength of public opinion on any emerging event or topic in Pakistan.