Create A Machine Learning Model - Based Around Student Retention

A proposal has not yet been provided

Hi, Below is my experience, and I can deliver you quality work with in specified or agreed time. Myself Ph.D. in advanced analytics having 10+ years of experience in developing and delivering analytical projects using open source R (including in-memory computing), and can deliver your requirement with R, accompanied by a step-by-step word document, what each line or code means. I have done this earlier using a kind of mixture model. Hence, have experience in it. As, freelancer.com bid does not allow for attaching; cannot showcase my expertise in this area. I am available at freelance chat (click on my name and options), to understand the requirements and explain my approach if required in detail. Waiting for your reply. Regards, Dr. Kumar PM.

New to freelancer but with good machine learning skills, confirmed data scientist and with a wide experience in predictive models using python and R, Hope we will work together .

Dear Sir or Madam, I am M.Sc. in Applied Mathematics (Germany). Currently I am a member of a Research Group in Computational Intelligence . My skills in Machine Learning: Clustering(SOM,K-Means, Neural Gas) and Classification Algorithms (GLVQ, GRLVQ,GMLVQ) Programming skills are listed in my profile. Also I have strong background in advanced statistics and theoretical computer science. Best Regards, Grigory Devadze

Dear Sir, I am a Master's student and my area of research is Machine learning. I had read the job description and I feel that I will be able to solve it. This is basically a binary classification problem. There are 18 attributes and 19000 data. 17000 data can be used as a training set and rest 2000 can be used a testing set to check the accuracy of the classifier. We can model a 18 dimensional feature space and use a linear discriminant analysis to analyse the separation of the data. In case the classifier turns out to be highly non-linear, we can use the kernel trick for better separation of the data. This is how I would like to initially approach the problem and analyse how it works, in case it doesn't works fine, I can move on to other advanced techniques. The best which comes to my mind now is by using sequential projection learning with hashing. I have worked on the semi supervised part before and I would say that it is a very powerful algorithm for classification. Kindly let me know if you think I can be the ideal candidate to solve this problem. Thanking you. Regards, Soumya

Hi, This is an interesting project. I have done several machine learning projects, mainly on python (pandas, sklearn and pylearn2), and I think the difficulty depends almost entirely on the data, so let me ask some questions: - In the attached doc is said that the best models are based on random forest and neural networks. Which metric are you using to validate the models? - what are the results you have obtained so far? - are there different datasets for testing purpose, or the results are obtained by crossvalidation? - it's hard to estimate the different issues that can arise (covariate shift, outliers, ...) without any dataset. Would you be able to provide some kind of anonymized dataset? Regarding the web application, I think it depends on the datasets also. For example, if there is covariate shift related to different universities/colleges, maybe different trainings must be done for each university, if there is enough data. Please, feel free to ask any questions you have. Best regards Raul Rios

So, you have come to the end of term and realized that you do not have the time to write your final dissertation or essay. Thats where Anacris comes to your rescue. We will provide you with an authentic, well researched and fully referenced assignment to get you the grade that you deserve. Strong analytics background in statistics, data mining, and optimization;Solid understanding of database systems, data acquisition pipelines, Unix scripting, data manipulation, data cleansing, and data management skills;Extensive independent consulting experience for profit and non-profit organizations;Results driven, decisive leader with success in strategic thinking and problem solving;

I am a Computational Biologist. I have advanced skills using R and Python and have a lot of experiences in machine learning and statistics using biological data. It's sound like that you get a nominal data set from the Universities. If yes, then I will go for rules and clustering. I can't tell approach without seeing the metadata and some infographics. Give me some Demo data then I can tell something. Please let me know. Thank You, Narendra Meena

Hi! Im an electrical and electronic masters student studying in UWE. I have the experience of coding in C/C++/Assembly/Arduino/PIC C/MikroC/MATLAB for more than three years. I also have the knowledge and experience of electrical and electronic circuit analysis and design. I have done BCS (British computer society) diploma level so I have the knowledge on PHP/HTML5/mySQL for website development. I have completed many freelancer assignments successfully. If you are interested in hiring me, please send me a message. I have made a lot of reports in the recent past. I have good experience in using Microsoft office softwares like Word, Excel and Powerpoint. I can write reports of any length without mistakes. Thank you!

Assuming each of the 19,000 rows represents a student profile, and the 18 columns are the attributes, we will first analyze the pattern of the attributes, and assign a weight to each attribute according to its relevance, then transform the attribute values into numeric forms so that each student profile becomes a vector in 18 dimensional space. Now given a profile P from the 19,000 profiles, we can calculate the distance between P and any P' from this same set, thus a machine learning model is built. Then with the actual drop-outs known from historical data, we can start training the system by fine-tuning the assigned attribute weights till an acceptable error threshold is achieved. Given a new student profile Q, we can pick out a number of K students from the set that have shortest distance to Q, and calculate the probability of Q to drop out.

A proposal has not yet been provided

I need to understand what models have been tried so far. Some feature engineering, exploratory data analysis may also be needed. I also need to understand what your goals are. Basically there are two kinds of model building goals: (1) Explanatory Models: Find models that best explain the variation in the data (2) Predictive Models: Develop models that have good accuracy. Ensemble models (like random forests) do well when there are disjoint regions of the predictor space with one model doing well in a particular region while doing poorly in others. The high accuracy is because you pick an appropriate model for each region of your predictor space. Nueral Networks also do well in the same setting. They are are also good when there is a lot interaction between predictors. So if you are satisfied with the accuracy but need actionable insights then the following might be useful: (1) Segment your population based on a discussion with a subject matter expert. If this insight/resource is not available, clustering your data will yield the sub-populations. (2) Develop models on the sub-populations, analyze errors and refine models till you arrive at a satisfactory model for your sub-population ( need to do this for each sub-population/cluster) (3) Analyze the feature importance for each model. This will yield your actionable insights

Some points which make me fit for the project 1. New to freelance and trying to explore different genres in analytics 2. Majority of my work is with customer segmentation and retention 3. Three years experience with R programming 4. Very good with machine learning algorithms and statistics 5. I can explain the results even to a least knowledgeable person in analytics I think it is a pure waste of time to attach my whole resume with it so I have given brief details so that you can decide how much you can expect from me !!!! Tools Used: 1. R 2. Tableau

A proposal has not yet been provided

I have consulting experience delivering this kind of projects. I already have some pre-developed scripts in probit (probability models) and other analysis that will make the project very efficient and easy to deliver.

I am a data scientist with more than 10 years of experience developing models using machine and statistic learning techniques . Below is the approach i will follow : a) Data Exploration : Understand the variables in terms of their distribution and correlation with respect to retention rates b) Data Transformation : Imputation of missing values if required and trend analysis c) Model Development : Since this is a small sample problem , I will probably use regression techniques with bootstrapping & cross validation rather than using machine learning like random forrest or boosting. d) Model Validation : The development will be done on 75% of the sample and I will keep 25% for validating the models to see if it will hold true for other dataset applications that you may have The above analysis will be done in R Please reach out to me for further questions .

Hello As long as the dataset is clean this should be a very easy classification model to build. My question is what format do you want this model delivered in? I can build the model and provide it in JSON PML, PMML, Python, Ruby, Objective-C, Java, C#, and R. If you plan to use the model on an ongoing basis I would recommend PMML format as this format is quite portable and may be used on and "Openscoring" server as an REST web service called to as needed. I primarily work in Rapidminer and can integrate data massaging macros written in R if need be. Thank you for your consideration. Mike

I have developed machine models which can automatically classify traffic as spam versus non spam. This is being used by a leading organisation. The requirement here is pretty similar and I can do it with a high level of accuracy. I shall be using R for developing this model. I am pretty sure you would prefer someone who is a hands on expert working in a leading company rather than anyone else.

Még nem kapott javaslatot

A proposal has not yet been provided