Posts

Additional Conclusion Post

Image
For a demo of my application, please see the link below.  For a conclusion to my AI project, there are a few things I would like to reflect on. The first would be the topic of lessons learned. Thinking about what I have learned as a result of this project, there are quite a few things that come to mind. The biggest single thing that over shadows the other lessons would be the topic of search algorithms. This was something I really did find fascinating during my AI class lectures. However, until I actually started to develop an application that could benefit from an AI search algorithm, I was unable to appreciate how complex, and important those concepts were. Also, after developing my project, I did find that most search algorithms work best for a tailored problem. For my project for example, a Monte Carlo search algorithm would be too complex, simply because that is tailed for huge data sets, and would be largely inefficient.      For the second topic I would like to ref

Lure Orical - Conclution

Image
In conclusion, after completing my final project, I have had some time to reflect on the process, challenges and results. The Process to design my Fishing Lure Orical application was to write something that I would use for my self. I felt that if I wrote something to satisfy a need that I, myself had, then other people would find it beneficial as well. I feel that the process of my thinking, how I choose to design the layout and the functionality of the application was close to spot on. It is easy to say that however, with myself being the target user. However, after speaking to several avid fisherman, they feel that this application would be something that will benefit them as well. In regards to the challenges of my project, I would say the 2 biggest challenges was gathering a large enough sample size and designing an algorithm that met not only my personal requirements, but the class requirements as well. The algorithm I choose could be viewed as simplistic, however at the tim

Orical Command Panel window

Image
For the final piece of my application, I made made the  Orical Command Panel window. This window is the equivalent to the home screen of a browser. Once the user opens the application, this will be the first window to appear. The user can then either select Add Fishing Log  to add fishing historical fishing experiences. Second, the user can select Lure Forecast  where the user can enter the fishing conditions they expect so the application can predict, based on the historical data entered by the user, the top 3 performing lures. This will allow for a more seamless experience, and help tie all the windows, options and data together so everything can be a better experience for the user. I have collected what I thought would be enough of a sample data from my self and other seasons fisherman. However the end result is, due to the bad weather in the past few months and due to my current algorithm implemented, the sample data isn't sufficient for meaningful lure predictions to be mad

Orical Results window

Image
For the results window for the lure suggestions, I have chosen the format displayed on the upper left hand side. Once the user enters the fishing conditions, the application will take that information and historical fishing data with matching conditions. It will first sort that data by lure that was chosen for that fishing attempt. Next, each of those lure categories are then divided into two sub categories: Successful and Fails. This will then result in producing the 4 most important values associated with that lure. 1) What is the total amount of fishing attempts made with that lure (that matches the fishing conditions provided by the user.) 2) What are the total amount of Successful fishing attempts completed with that lure (that matches the fishing conditions provided by the user.) 3) What are the total amount of Failed fishing attempts completed with that lure (that matches the fishing conditions provided by the user.) 4) Finally the most important metric produced is the lu

Current Conditions window

Image
For final section of my project, user will be able to select an option to allow for the application to suggest lure based on the current conditions. The user will be promoted to enter Air Tempature, Wind Speed, Sky Conditions, and Water Clarity. For Air Tempature, the options will be: "Less Than 60", "60 - 75" and "< 75". For the Sky Conditions field, the options to select will be: "Clear", "Cloudy", "Rain/Mist". For the Wind Speed field, the options the user will be able to select will be: "Calm", "Moderate" and "Windy". For the final section Water Clarity, the options the user will be able to select will be: "Dirty", "Visible > 3 feet", "Clean". Once the user has completed these 4 fields, they will be able to submit this information. The application will then find previous historical fishing experiences that match the provided criteria and select the 3

Front-End Application Development

Image
For this portion of my project, I get to play with my favorite area of Computer Science...Front-End Application Development! There is always something appealing to be able to take a blank canvas and paint whatever you want on it, and then to give that paint functionality. For this portion of my project, I have built from scratch a user interface that takes data from the fisherman, and logs it in a specified format so it can be mined at a later date. The interface for this project does not require any thing fancy, however I do enjoy making something sharp and appealing to the eyes. For this section it is simply asking the fisherman for the weather, water and fishing techniques that were present and the time one single fishing attempt was made. A fishing attempt for this project is defined as Fishing for 5 minutes with the same lure from shore. This allows for consistency, so the integrity of the data is not distorted, and accurate predictions can be made. F

Project Update

To help solve the problem of lacking data, I have enlisted several seasoned fisherman to help log their catches and send me the relevant data. This has helped some what, however just as with most data mining, there's never truly ENOUGH data. At this point I am at 10 ACTUAL logged fish catches. This is defined as, a fish was caught this year using the defined specifications (must be fished from shore, using the same lure for less than 5 minutes, ext) and all relevant data was captured. The main reason for the lack of successful fish capture data points is the weather. It has been unseasonably cold this year, and especially due to the recent blizzard a few weeks ago, fishing has became very difficult. The only "reasonable" way to solve this issue by the dead line is to use historical data from last year. This data would be coming from pictures of fish caught by my self last year, which I can remember the lure and conditions which the fish was caught in. I feel that this met