HelpIsHere

Inspiration

This year (2017) has been labeled as the period with the most devastating natural disasters. Fifteen separate weather and climate disasters have each caused at least $1 billion in damages in the U.S., according to the National Oceanic and Atmospheric Administration.

This unite struggle gave rise to the idea of the “Help Is Here” application (the HIH app). The goal of this app is to be used globally for maximum effects of providing the necessary assistance to those in need.

Many non-profit organisations and self motivated volunteers initiated events during the year of 2017 to help the people that suffered from destruction such as the hurricanes in Florida but what both committees lacked was the immediate response. Our app will allow these organisations to expand and use all the helping hands they can.

People usually want to help but are discouraged by the lack of motivation to join the organisations our app will also solve that problem because they will join in the comfort of their home by joining the app and then it will be up to them to respond when called.

What it does

Our app will have two options:

Helpers /NGO

This option will allow everybody to sign in and create a profile with their location, age, religion, ethnic preferences, level of comfort dealing different situations, contact preference and the qualification of the individual. This profile will later be used to match the helpers with their PeopleInNeed.

PeopleInNeed

Another option will be the PeopleInNeed else called PIN which is going to be a shorter survey of what type of need these people require. Matching them will be a matter of seconds for our app and the helper will be notified immediately then the volunteers or organizations will have contact information of probable hosts for the refugees or certified first response individuals for medical emergencies expanding the resources provided for all by all.

The Matching Algorithm

The matching: DIET is an algorithm which uses a simple wrapper approach to heuristically search through a set of weights used for nearest neighbor classification. DIET sometimes causes features to lose weight, sometimes to gain weight and sometimes to remain the same.

DIET Algorithm

In the DIET algorithm we have a discrete, finite set of weights instead of continuous weights. If we choose k number of weights then the set of weights will be: {0,1/k,2/k,...,(k-1)/k,1} If k = 2, then the set of weights would be {0, 1} which means that we either give weight = 0 or 1 to an attribute. When k = 1, we have only one weight which is taken as 0. This translates into simply ignoring all the weights and predicting the most frequent class. Generally when we have k weights, we start with the assignment closest to the middle weight. For each attribute we move through the weight space in search of the weight which minimizes the error until minimum or maximum of the weight is reached. The number of neighbors used in the classification is 1 since the goal is to investigate feature weighting rather than the number of neighbors Error is calculated every time using tenfold cross validation over the training data with KNN algorithm. A halting criterion is used where in we stop the search when five consecutive nodes have children with no better results than their parents. (0.1%)

DIET Algorithm: Weighted K Nearest Neighbor Alorithms

Challenges

Technology unfortunately is not accessible to all
Users may have strict preferences. Fulfilling everybody’s needs while sustaining a safe and inclusive environment is hard to be coded with math.

Check out the Work Flow: coggle-HiH