Data Logging - Science topic

Dear Safdar:

I agree with previous answers view of point, and i confirm that issue is related to specifications of the work platform you used, represented by the processor and GPU.

Thank you for your constant help Hayder Dibs . Garmin GPS will be costlier for us, because we want to deploy those on fishing vessels and there is hardly any network out in the sea, so I doubt how effectively will cellphone GPS work.

Thank you but sorry for the late response, Dudley. I will update the timestamp and the furnace performance. Wll update you soon on the results.

If the normal distribution is a bad model for the response, is there a more approprioate model? If so, then you should use the more approriate model.

What such model could be depends on what your response variable is (you did not mention this - but this is the MOST IMPORTANT information about your experiment!). If these are counts (e.g. cell counts), you should use a negative binomial model, if these are concentrations or intensity measures, you should use a log-normal or a gamma model.

But still it may be that the variances are too large to see systematic differences in such relatively small samples.

Excluding outliers because they don't suit your hopes is a no-go. Checking for outliers just helps to identify *possible* problems with samples, so that you can go back and find a reason why there samples might not be trusted. If you don't find such reasons, the outlier provides the very same information like any other value and is possibly the one important observation telling you that the rest is a bit biased.

Often, outliers seem problematic just because you have a particlar (and possibly wery wrong) intuition about how the distribution should look like. Changing your distributional assumptions may make these values way less suspicipous (there is a phrase: "I am not an outlier! - I just haven't found my distribution yet.")

Look at here

http://chemequ.ru/%D1%81%D0%BF%D1%80%D0%B0%D0%B2%D0%BE%D1%87%D0%BD%D0%B8%D0%BA%D0%B8/%D0%BA%D0%BE%D0%BD%D1%81%D1%82%D0%B0%D0%BD%D1%82%D1%8B_%D0%BA%D0%BE%D0%BC%D0%BF%D0%BB%D0%B5%D0%BA%D1%81%D0%BE%D0%BE%D0%B1%D1%80%D0%B0%D0%B7%D0%BE%D0%B2%D0%B0%D0%BD%D0%B8%D1%8F/

Good luck

Hi Blessings Hendrina Likoswe

This is how I interpret your results.

1. The distribution of errors, with and without log transformation, is not normal. The histograms are usually misleading as what appears to be normal may turn out to be non-normal if we change the number of bins used to plot the histogram. Similarly, the tests for normality are also misleading as they are usually significant if the dataset is large enough. The QQ-plots are a good way to check the normality.

2. I would argue not to log transform the data simply because then you have to live in the log world and the units of measurement become log-units. If they have intuitive interpretation in your case, go ahead and log transform otherwise not.

3. From the Altman plots, a few conclusions can be drawn. There is a systematic bias in your measurement method 1 compared to 2, or the way around as you don't have a gold standard. This is, however, not the end of the road. This systematic bias can be accounted for by regression analysis which is reported as "y(differences) = -0.027 x(means) + -0.35" in the non-transformed data report. This can be interpreted as the systematic difference between the two methods at a measured value (means stand in for the true score).

4. As the errors are not distributed normally around the systematic trend (or the expected value of the differences), the assumption that there is a constant LOA around the systematic difference is null and void. LOA is computed as bias + 1.96 (z-score from standard normal distribution) * standard deviations of differences. The violation of this assumption implies that in a practical situation, you may expect smaller random error at smaller measured values and larger random errors in larger measured values, once the systematic error in the instrument has been accounted for. You still need to interpret the size of the error based on the application of the measurement methods.

My conclusion is that the reliability of the methods seems to be less than desirable. The methods may still be applicable with a mechanism to account for systematic errors and the random error is expected to be large for larger measurements.

thankyou very much!

If you use just a classical thermocouple, you can easily get out the cable using specified feedthroughs, e.g. https://www.vacuum-shop.com/category/2073666/thermocouple-feedthrough-flanged-type-k.html (but check for cheaper competitors, of course). Then you can just wire it to any cheap multimeter and read in the values via an RS232 that every decent computer should have. Then you don't have to drill extra holes and you don't have to worry about tools not working inside vacuum. I've actually never seen anyone put the logger itself into vacuum.

AVAILABLE IN MARKET

We recently used PXM409

Hi Vytas, Yes indeed, it was that package I've been talking about. However, since this is an old question, I managed to solve the problem - even though probably not in the most straight forward way due to a bunch of error messages I needed to bypass.

Hence unfortunately, I can't provide any good explanations in how I did for others experiencing the same issues.

But cheers for the script, I will let some data run through it.

Cheers, Claudia

Hello,

More cost effective solution would be __any__ two (current + voltage) digital multimeters equipped with a digital interface. The ranges of such meters should meet mesuring PV system constraints. The cost of such measurment system development would be in this case moved towards a time spent to manage the process of acqisition (and storge) the measurements results onto a PC via a digital interface.

I hope you will consider this direction useful.

Adam.

Its is really depending on the production line.

The most important procedure to prevent constantly break downs is by doing the periodic or predictive maintenance...when you do that you could prevent better the sudden brake downs.

Second thing is the quality inspection at each production step ... this will let you know when the error starts....the measuring of these errors is due to products.

Searching for the problem start on different levels:

- Problems in power supplies

-change of the rough material (defects is raw material)

-small Details in the Product Design may cause brake downs after certain time of Production.

-Cooling or heating operations during Production maybe a big deal.

-and obviously malfunction of the equipments this one can easily find.

Use the build matrix function together with the write to spreadsheet file VI.

It looks as though your hardware is set up correctly: At least you have a resistor across the data logger inputs, which effectively converts the 4-20 mA current to a voltage, which is what the data logger expects. Check your DT80's settings: It may be reporting an integer number of millivolts or something similar. Data loggers do not necessarily report readings in "engineering units" (e.g., PSI, volts, etc.) if you don't tell them what to do. The DT80 may be doing exactly what it should be doing, though I presume you can program it to convert inputs into engineering units using linear math functions (I'm not familiar with the DT80's details). How do the numbers change as you apply pressure to the pressure sensor? It is likely that your numbers are varying because you're seeing natural "noise" around a small signal (your three numbers enable us to calculate that the standard deviation is only ~4% of the mean; not bad for a signal near the noise floor). If you see your numbers increasing with increasing pressure (1 kPa is not very much- don't apply too much pressure!), everything may be well. You may have to develop a regression between data logger readings in millivolts and pressure in kPa.

Update: To explore an interaction in logistic regression I have seen recommended a simple chi square test, splitting the data of one IV to examine the effect of the other IV. I have done this and it all makes sense.

I used 4 Lascar dataloggers ,(https://www.lascarelectronics.com/easylog-data-logger-el-usb-2/)  in the humid montane forest of the Andes. However, two of them were dropped in a ravine and got ruined because they were covered with water. They measured temperature and humidity every hour for 6 months. However, they can't measure precipitation. I do not know of any datalogger that can measure that.

You might also check out Onset Hobo data loggers. The old versions several decades ago were very inexpensive, and needed to be contained in a water proof capsule.  To save funds, extra protection was added with plastic whirl-pak bags as very inexpensive. The new data loggers are likely better and more expensive, and some issues with people finding and damaging, so se started to include a note on who to contact, what the temperature recorders were doing to help identifymand improve fish habitat.  I have had good results with them (Hobos) as long as not damaged,,stolen or taking on water.  The technology has improved in the last 20 years.  I have not used Tinytag, but in an earlier researchgate question, some individuals found them reliable.  I made sure we secured them in sections of plastic pipe that were cabled to strong roots or other relatively immovable oblects, hand burying cable under rocks if available to obscure detection.  It is important to provide map and description of location if the person installing will not be servicing the data loggers, perhaps even capture a picture of site.  If stream is likely to be intermittent, you may want to locate in pool, but I usually preferred sections with more velocity to avoid sediment burial if evident.

OK, so you might want to have a bluetooth connection between Edison and the mobile. And some WiFi or xG from the mobile to the 'cloud'.

So you cold implement an HTTP server on Edison which would be polled by the client part of the mobile app.

The 'cloud' side is trickier: as  most likely you will not find a 'cloud client' polling your mobile app. Thus you will have to find out how to actively transmit the data from the mobile app to some receiving service in the 'cloud'.

'cloud dashboard' ? What's that ?

Don't you have some colleague eventually showing you what's already available (in terms of concepts, not necessarily 'implementation') ?

Regards

There is some work that has been used in classroom environments. This work has relied on log files, human observation, web-cams, other sensors, or some combination of these inputs to detect student affect. 

Pardos, Z. A., Baker, R. S. J. D., San Pedro, M. O. C. Z., Gowda, S. M., & Gowda, S. M. (2013). Affective states and state tests: investigating how affect throughout the school year predicts end of year learning outcomes (p. 117). ACM Press. http://doi.org/10.1145/2460296.2460320

Bosch, N., D’Mello, S., Baker, R., Ocumpaugh, J., Shute, V., Ventura, M., … Zhao, W. (2015). Automatic Detection of Learning-Centered Affective States in the Wild. In IUI 2015 (pp. 379–388). Atlanta, Georgia, USA: ACM Press. http://doi.org/10.1145/2678025.2701397

A self-reporting method called experience sampling can also be used and combined with other data collection methods to reduce participant burden or build affect detectors so that you can reduce or remove your dependence on self-reporting:

Clayton Epp, Michael Lippold, and Regan L. Mandryk. 2011. Identifying emotional states using keystroke dynamics. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '11). ACM, New York, NY, USA, 715-724. DOI=http://dx.doi.org/10.1145/1978942.1979046

Carrie Demmans Epp. 2016. English language learner experiences of formal and informal learning environments. In Proceedings of the Sixth International Conference on Learning Analytics & Knowledge (LAK '16). ACM, New York, NY, USA, 231-235. DOI=http://dx.doi.org/10.1145/2883851.2883896

When I used experience sampling, 11 of my 12 participants submitted enough reports for me to use the data. The bigger challenge with this approach is getting people to start the study. If you go the self-reporting route then I would recommend combining it with another sensor or data collection method. There has been discussion about using smartwatches or fitbits to support affect detection but I haven't seen any papers on that yet.

Hi Daniel,

It all depends on which type of habitat you want to use them and how long you want them to be in place.

We have used many of them recently in rivers for recording temperature and lgiht during a whole year. Great for temperature, not so good for light (lux) as the case can get eroded (suspended sediments), you get biofilm growth, or you can lose the whole thing because of floods. In our case we attached them to bedrock on the stream banks with steel cable and a weight so that they would stay in the benthos. Our recovery rate was high (up to 90% of them), however some of them were out of the water when we collected them (less than 5%).

In other aquatic systems like ponds, lakes or other lentic habitats or really stable streams (spring-fed) this might not be an issue, however vandalism might be another problem!!

I think the greatest challenge is to keep them measuring light always in the same angle and position. If you are really interested on measuring light in different seasons you might need to leave them for short periods when the flow is stable and devise a way (e.g., attach them to a heavy flat tile which is fixed to a bedrock or big boulder with steel wire) to place them always in the shame location and position. This would apply for lentic and lotic habitats, I believe.. Another thing you might need to think of is the position of the sensor within the aquatic habitat, as light intensity could be spatially heterogeneous. Replicated locations might be an option..

Cleaning them will defenitively be a must if you leave them for a long period!!

If you get all that going I think the sensors are of reasonable good quality for their price. We also use them when we measure river metabolism for short stable periods and they correlate quite well with other more precise light sensors.

I hope this helps!

Good luck!!

Hi Andreas, I am interested in the interaction between play and emotions myself. The current focus is gamification rather than "real" games, but still the attached paper might provide a few hints.

We used facial recognition to get a more objective view on a person's emotions than the usual questionnaire can provide. However, I feel that this is just a starting point. Multiple sensors are required to consolidate this kind of physiological research. I also got the impression that current systems (like the SHORE-kit for facial recognition we used) have been developed for lab conditions and are often difficult to use in the field.

Maybe we can work together on this topic. I'll soon start a new project on measuring emotions in the context of (gamified) learning processes.

Arduino is easily capable of delivering 10mA directly, as single pin can draw/sink 20mA. the problem lays in proper voltage level. If you have 3.3V at your board - use it as power to your sensor!

The other issue is reference voltage. If you use Arduino Mega, you can use internal 2.56V reference with analogReference(INTERNAL2V56) command (please take a look on https://www.arduino.cc/en/Reference/AnalogReference)

This will give you readout values :

0% - ~131

100% - ~530

However, if you are interested in even higher resolution, you may use analogReference(INTERNAL1V1), which will clip your measurement range down to 76%, but you will get resolution like:

0% - ~307

76% - 1023

I do not understand 6the quastion even

Hello,

you've still gotten some work to do :)

the problem is that I know nothing about YOUR odometer and especially its speed sensor. Typically 2 types of speed sensors are 'standard': tacho generator (primarily delivering an AC voltage whose amplitude is proportional to the speed) and 'pulses' - meaning that you get 1 or more pulses per round.

Theses pulses may come from a simple reed switch (as with bicycle odometers) or from some kind of electronic sensor (normally a an inductive sensor or a hall sensor or something alike if talking about odometers from cars). Talking about cars the odometer may not even attach to the sensor itself - simply processing a signal generated nowadays by the ABS unit and delivered via CAN.

So, the first thing is to find out which type of speed sensor you've got. This means tapping the sensor interface (preferably with an oscilloscope) and monitoring the system during normal operation. Talking about pulse output sensors you may find out that the signal looks somehow 'disturbed' - meaning that the signal is missing pulses on a regular basis. This is normal for a lot of automotive sensors.

Not to handle too many alternatives in this reply: please come back when it's clear which type of speed sensor you've got - preferably with some kind of schematics and measurement plots. (If you can get documentation for the odometer this would also help.) Then it will be easier to assess how to proceed.

Hi,

I do not know the odometer system, but as long as it can send out an electric signal then it is possible to log the signal inside the data logger. I had an experience to log solar irradiation into a computer by means of ATMega8. Take a look at the attachment.

Good luck

Of course. First off, you need to know what communication protocol the instrument talks to the computer. If you wanna use the Matlab code you need to connect the instrument using ethernet cable. If it is GPIB or Serial COM you would have to change the code accordingly, depending on what VISA resource that is. Second, the code is just simply a function, you have to type in your command into the terminal to be able to read stuff from it. This function applies voltage sweep from one channel and forcing the voltage to be zero from the other channel and measure the current from that exact channel. If you want to do some stuff other than this, you have to modify this code accordingly as well.

Personally I have never worked with Thurlby multimeter. And I can't seem to find Keithley 1906 (I don't think they sell this....)check 2400 maybe?

For a one or two unit system, definitely an arduino based logger. 

If you need a *bunch* of sensors, it is probably worth doing a custom board and having the whole thing fabbed overseas. 

I'm looking at a project that needs 10k individual sensor/logger units, and with those kind of numbers, you can get per unit cost (~8 gb per-unit storage) down to around $40 a pop. They're not the worlds greatest loggers, but not terrible. At all. 

Hello Ronny, I usually prefer to log sensors on a logger, but if you look for a really portable, handy instrument to deploy in few minutes, the Kestrel 4500 Weather Meter is outstanding: http://kestrelmeters.com/products/kestrel-4500-weather-meter

Not cheap, but sturdy, reliable and rich in many features.

'Hope it helps.

Best, Andrea