Both tanks synch

SFS tank 01

Refuellings

See the flare out of the 3 compartments at the end. Why?

Reco betw Omni and Hectronic

Reconciliation of refuellings

Jajang

First fuel modelling of Jajang Dumper

Two weeks data model

Updated till 1:40 pm

SHows hourly data where > 0 litres consumed

First fuel level upload for verification

Fuel level dipping for several minutes not reaching zero

two small refills

Now we can see the section of fuel graph with low EER (high CPH).

Drain detection POC

leakage in moving asset

6 to 8 am

This one has peculiar spikes in raw levels.

A noisy sensor example

Balance 42 assets were tested today for 3 dates.

The blue dots are after applying tsclean() ... green lines are original data so here tsoutliers has worked perfectly to detect the outlier points at 02:14

The blue and green lines are coinciding which means tsoutliers has failed to detect the outlier points at 02:14

Zoom into the May 10th date to see the clear drain

Just before the filling starts there is huge fall in level

7 am to 8 am consumption is very high

Is this fake refuel?

From 1500 to 1600 the consumption increased 5 times normal

Sensor seems dies at 1 AM.

all vehicles bird's eye view

The detected drain by omnicomm is 22 lits while actually this looks to be 42 lits drain.

is this a drain test?

POC base reports

These are analytics on APCO vehicles

base reports produced from the POC

Fuel tsd facet ncol =1

Two key operator performance on same assets

EB Watts, DG Watts, DCV and Ibatt are the 4 dimensions on which we use a kmeans with k = 5.

Over two window sizes: 365 days and 1458 days

To get a feel of variations in the distributions of energy across the POC sites. ALso across dates for the same site.

Classification of Time Series Data of battery voltage into three classes using Decision Tree Learning.

Bagru2 has many ELM errors.

Dud DG performance

Kalwara DG performance analysis

Dand Analysis of LPU

WIth both battery and DG on shaded

Reliability measured as count of observations received every 15 minutes across all parameters from Ganapathi plaza from ELM4100 DC EM

Observation counts of ELM8420 in DAND for the entire day of 22nd, captured every 15 minutes.

These are number of observations received every 15 minutes for each parameter seen separately. The data shows RUT955 counts for the entire day of March 22 (midnight to midnight).

BNP field survey results summarised

Except Roads, Drains and Footpaths all other expense heads included in the per capita expense histogram.

Boxed wards on both sides. X-axis is Per km spent on Roads, Drains, Footpath and Lighting.

Sample waffle charts for Bellanduru budget utlization. Also shows relative size of budget across each year

Designing this for social media

Histogram across all wards on Toilet Budgets

Histogram of wards showing Budget per capita on Lightings - highlighting bellandur

Hisotgram across all wards Budget per capita on head: parks

Drinking water per capita histogram across ward

Histogram of Roads, Drains and Foothpaths - Budget per capita

Scatter plot for all 198 wards, total expense vs total road length. In addition color of point denotes: Political Party. Size: ex Mayor (large size) In addition we see the fitted curve & correlation coeff.

Normalised expenses means Expenses per km or road existing in the ward.

Unit expenses (Expenses per kms) on roads

test

BBMP Budget for FY 2019-20, ward wise

Expenses over 2015 to 2020

Utlisation of BBMP Budget from 2015-16 to 2019-20

Projects between the value of 10L to 1 Crores and who have data stamp availability for FINISHED DATE and PAID DATE in all 4 combinations. The points are shown in jitter mode to expose the density of the projects every year.

Expense to Approved amounts

Sample of 10 wards plotted on Net Spent year on year vs approved amounts on the same job numbers that were undertaken.

We are using this to test our citation function

for checking attendance overlaps

Testing the publishing

Shift duty rosters can be published using spreadsheets but then the actual IN and OUT of an employee is the difficult thing to track - even if they are electronically checked in and out. The data is immense and difficult to visualise in relation to each other. These duty hours line charts are simple to understand and give immediate insights.

Seems a hack attempt from Macau

these mysql logs include all SELECT statements of jupiter along with APIs that start around 22:44

these logs are combined logs for 10 minutes of API firing as well as MySQL: including ejabberd and jupiter_dev

These logs have all UPDATE, INSERT including XMPP MySQL,. Note: We are truncating the scripts at 40 char.

API log merged with MySQL log

filtered on mysql

last 100 queries

Listing of topics and subtopics in class id 22

R API flooding

These are sample logs captured

with color text

TIme plot of API counts

This doc explains when you are capturing data and you know you are going to grow a variable into many levels, how do you maintain the old factors.

just a simple histogram