How to plan your project
What you need for your reports
In your reports, you will need to provide some form of table or chart that shows when you plan to complete each `feature’.
I’m assuming here that you have already convinced yourself to take an agile approach to the work, which means that you start with a backlog (or list) of features. Your list of features might be something like this:
Feature |
---|
RNN model |
Data cleaning |
Hyperparameter search |
LSTM version |
Evaluation strategy |
You need to make sure that there is enough detail in the description that you know what each item in the list refers to. You will then need to organise the feature list into a priority order. Usually, you want to do basic or generic things, like work out your evaluation strategy, first. So a revised feature list might look like this:
Feature | Relies on |
---|---|
Evaluation strategy | |
Data cleaning | |
RNN model | clean data / eval |
LSTM version | ” |
Hyperparameter search | rnn and lstm |
I added a “Relies on” column so that you can see the reasoning behind this ordering.
The next stage is to work out how much time you have and to make a table with a column for the week beginning date. These will be your sprints, to use the agile term for them.
When | What |
---|---|
12-7 | Evaluation strategy, data cleaning |
19-7 | RNN model |
26-7 | LSTM model |
2-8 | Hyperparameter search |
Note that I have not left time in here for reading the literature and writing up your results. That’s because you should incorporate such reading and writing into each sprint. Here are some of the things that need to go into a sprint:
How much | What |
---|---|
20% | Literature review |
20% | Write what you plan to do |
40% | Do it |
20% | Write about what you actually did |
You may be tempted to produce more glamorous project Gantt charts for your final document. However, keep in mind that these are generally just gloss and don’t provide any more information than a simple table with headings of When and What.
What you need when you meet with your supervisor
When you meet with your supervisor, it’s a good idea to have your plan at hand. In the case that you are ahead of schedule, your supervisor will be encouraged to give you ideas about how to deepen the study and mine the available `bonus’ points for your thesis. However, if things are not going well or perhaps need to be done in a different order, your supervisor might be able to help identify how to short circuit problems or get something done more quickly. You may feel some difficulty in admitting that things aren’t going well but be assured that it is quite normal for some problems to be encountered in any study.