Hello all
Happy new year!
I am Ajit Jaokar of team Miami Young Data Scientists
Great to be here and look forward to sharing and collaborating
we have some questions/clarifications
here goes!
Questions
the 15K data is for all the 9 sensors. yes? Is there a list of the actual sensors accessible(ie not all sensors in kit/ github ie those that are used in space)
whats the best place to get the technical details of the data and sensors? (re hypothesis below) I know the github but more looking for the hardware specs
How are people using the kits? (in schools) we have the kits and we were thinking of getting the team (participants) experimenting with specific experiments (same we are running in space) but welcome comments. ex we can take readings on the ground and use these as reference points so the students can get an idea of the kind of data they will get from space. Thats one way …
Hypothesis
I envisage below. Any feedback/questions/comments welcome
Overall our experiment is to predict conditions on earth based on data from sensors. This was mentioned in the call / webinar also … but in our case, we are attempting to use Predictive algorithms on the data to extrapolate i.e. make future predictions.
Hypothesis questions
a) Is there access to a space expert to cross validate ideas i.e. someone who knows space? for example What can be predicted(in terms of space) - ex clouds, thunderstorms etc (and by extension - what data is needed for those predictions) - how can I find this out/confirm from someone who know space(these may be some of the references in our work also - so we could propose the hypothesis - but great if we know someone who can say if we are on the right track overall … )
b) This is more an observation rather than a question. I don’t think we will have enough data to make a ‘proper’ prediction - so that’s why I want to know what can be predicted. Once I know that we can make the analysis and create a scientific method i.e. say that our results are based on the datasets we observed … etc etc.
c) This means my experiment itself will be simple i.e. just recording data from some sensors (based on point a above i.e. what is to be predicted). We may not need to run code but … we may have to if we have to programatically manage data.
d) Out of curiosity, what code are people planning to run in space and why? i.e. in our case, I envisage that the data science code would be run after we get data from space(in iPython notebook) which is Python code (vs Arduino code)
PS
Currently we are using this reference https://www.ardusat.com/lessons/50 but we still have questions specifically - but its a good reference to start
Is the 15kb data for all 9 sensors?
Yes, the 15kb data is the payload size for all the data that you plan on collecting over the course of your experiment.
Where to get technical details about sensors?
See Sensors Available for Space Experiments for information on the sensors that are available for use in experiments and links to their datasheets, which contain more detailed sensor specifications.
How are people using the kits?
Prototyping the experiments that they plan to run in space and using the sensors to get measurement of the environment here on Earth to get a sense of what each sensor measurement looks like is a great way to prep for space while learning important computer science and electronics skills. We also have a number of suggested exercises posted at https://www.ardusat.com/lessons (click on “Experiments” and “Missions” for activities designed to be completed with the Space Kit).
Hello Ajit, thanks for the questions! see answers below…
a) If you have a specific hypothesis question that you would like some feedback on, please submit and we will do our best to answer and send to our space partners if needed. Some of the questions may not have solid yes/no answers (and this also can take the fun out of experimenting) as many of these experiment ideas will be new for our specific payload. Regarding weather data, I really like the Dark Sky Forcast for global weather data by Lat/Long: https://developer.forecast.io/
b) I like the idea of building a predictive model. I agree that it may be hard to gather enough training data within the data payload limits.
c) Yes, it should be simple to setup your experiment and we foresee the vast majority of teams not requiring writing code.
d) The majority of our experiment requests only require data collection and like you mention teams may use code for working on this data on earth. We are big fans of Python, a perfect pick for your data science needs. We will be delivering datasets to teams in the Experiment Platform and there is a one-click integration that easily moves datasets into a Jupyter notebook data frame. In the future we are looking to expand on satellite code options that students may use to design programs that sample data in space and only use these values as they build their data payload.