Jns2183
-
Posts
4,927 -
Joined
-
Last visited
Content Type
Profiles
Blogs
Forums
American Weather
Media Demo
Store
Gallery
Everything posted by Jns2183
-
I'm made we got skipped over by storms once more Sent from my SM-G970U using Tapatalk
-
I think this has to be related to the theoretical max temp versus our theoretical min temp. Basically the closest you have to each extreme the more rare it is. But also seems like you're working with different statistical distributions for the high end versus the low end Sent from my SM-G970U using Tapatalk
-
That's a great question and you are probably correct. The data set I have downloaded is KMDT from 1/1/1941 through 12/31/2023. The question I have is do you want to know what the probability of 13 below normal days are for a rolling 20 year. Referencing the mean during each rolling 20-year period or do you want to reference the mean for the entire data set. Basically if the average July mean for entire set is 86, and the mean for July 1950 - July 1970 is 84, the 13 BN will be different temps Sent from my SM-G970U using Tapatalk
-
What are you talking about in reference to 60% more to 400% more? Sent from my SM-G970U using Tapatalk
-
For it not to be skewed, at minimum, The median would have to equal the mean but that's not reality. Sent from my SM-G970U using Tapatalk
-
But it has to be skewed due to the whole distribution being skewed Sent from my SM-G970U using Tapatalk
-
See I think the average Joe is going to care much more about a 13 below normal day in January or a 13 above normal day in July then they will for a 13AN in January or 13 BN in July. The numbers don't matter it's how it feels to them. I remember you were asking before why new stations don't make such a big deal out of days with a much bigger above normal spread than we currently have when it's winter. And I see the answer that is because it feels good in winter versus it feeling not good in summer with a much smaller AN Sent from my SM-G970U using Tapatalk
-
Haha. I went to Penn State, finished up at Penn State Harrisburg. I spent 2 1/2 years in Engineering before switching to Accounting. When I graduated I was one math course short of a minor, but took several upper level stat courses. I went back to bartending because I loved it and I could after my brother sold a business I helped him start. Before that I worked for government as corporate tax auditor dealing with complex financial instruments, tax shelters, etc. I enjoyed the intellectual aspect of my work but the bureaucracy really suck the living soul out of me and made me hate getting up every morning so I jumped at the chance of being able to not do that anymore. Are you statistics in figuring out what to even look for, and it's worth looking at, and in the understanding of the financial instruments themselves. Although numbers took a back seat to trying to figure out the legal framework and how to apply different things. Feel like I know enough about statistics to get myself in trouble and make myself dangerous by deluding myself. It's very difficult trying to put some of these concepts in simple terms and actually I have found out that some of the best teachers out there have been from Khan academy and some other amazing YouTubers. The biggest hurtle I've had was my resistance to learning python and working with big data and although I'm still just an amateur with it I'm slowly getting better. Chat GPT has been an absolutely amazing resource for learning how to apply Python to these data sets Sent from my SM-G970U using Tapatalk
-
First issue is that standard deviation is the language of a normal distribution which we do not have here. More likely we have either a gamma or about normal one or some Frankenstein like combination of the two. I will try to plug all the numbers in next week or so see what it comes up with for the distribution and following equation. That should allow me to produce a charge that roughly states what AN value a specific BN temperature corresponds to according to a relative frequency in relation to a mean temperature (most likely mean January temperature). In a short, in a real rough generation, 1 std dev will have different values depending on whether it is BN/AN Sent from my SM-G970U using Tapatalk
-
It's not that no one cares it's just about the rarity of events. I'm sure +20 in winter is order a backlitude some more likely than -20 temperature differential Sent from my SM-G970U using Tapatalk
-
Sure we can create a chart that uses the equivalent probabilities instead of temperature differential Sent from my SM-G970U using Tapatalk
-
@Bubbler86 this is the answer to your question of why no one cares when we have a plus 15 temperature differential in winter time. Because essentially it's three to five times harder have a plus 15 temperature differential and summer or a -15 temperature differential in winter than the inverse for each season Sent from my SM-G970U using Tapatalk
-
Officially hit 102 at KCXY Sent from my SM-G970U using Tapatalk
-
That's because temperatures don't follow a normal distribution. This asymmetry suggests a left-skewed distribution, where lower temperatures are more frequent than higher temperatures. A possible statistical distribution for this kind of data could be: Log-Normal Distribution A log-normal distribution is suitable for data that are positively skewed. It is used when data can be thought of as the exponential of a normally distributed variable. Gamma Distribution The gamma distribution is another option, which is flexible for modeling skewed data and can accommodate the higher frequency of lower temperatures and the occasional high-temperature values. Weibull Distribution The Weibull distribution is useful for modeling skewed data and can be adjusted to fit the asymmetry observed in the temperature data. Empirical Distribution Given the specific frequencies of certain temperatures, an empirical or histogram-based approach could be the most accurate. This would involve constructing a probability distribution directly from the observed frequencies without assuming an underlying theoretical distribution. To summarize, based on the provided data and the observed asymmetry, a log-normal distribution or a gamma distribution would likely be more appropriate than a normal distribution for modeling the temperatures in Harrisburg in July. However, for the most accurate representation, an empirical distribution based on the actual observed data might be the best approach. Sent from my SM-G970U using Tapatalk
-
101/63 Sent from my SM-G970U using Tapatalk
-
Bad bad news Sent from my SM-G970U using Tapatalk
-
Can you explain why we tend to get up so close to our high by noon time and then barely move from noon to 5:00 p.m. even though it's max heating Sent from my SM-G970U using Tapatalk
-
Here is KCXY Sent from my SM-G970U using Tapatalk
-
Here is KMDT Sent from my SM-G970U using Tapatalk
-
99/68 KMDT is 97/70 Sent from my SM-G970U using Tapatalk
-
Stop in for a drink my drought brother. I'll give you a top shelf cocktail for $8 Sent from my SM-G970U using Tapatalk
-
I took screenshots of the 5 minutes or the observations for Capital City airport and for Harrisburg international airport. The first picture of some capital city the second picture is from Harrisburg international. These are all from last night. Sent from my SM-G970U using Tapatalk
-
What direction was the wind blowing Sent from my SM-G970U using Tapatalk
-
Do you live near a creek, marshland, a swamp, corn fields, any agriculture? Your dew point is always so freaking high Sent from my SM-G970U using Tapatalk
-
Maybe if they hit a 100 it unlocks lightning. It's so weird seeing these storms with a thousand lightning strikes an hour ram through northern pa while we can barely manage to get a storm to not collapse within 20 minutes and produce 5 strikes Sent from my SM-G970U using Tapatalk
